New Ekonomistas post. This is an English translation.

As is well known, the Riksbank frequently publishes a figure of the debt-to-income ratio, that is, household debt as a percentage of disposable income. But the debt-to-income ratio is an unsuitable risk measure, since it at a closer look hardly gives any information about any risks with household debt. Among aggregate risk measures, the *interest-to-income ratio*, that is, household interest payments as a percentage of disposable income, is a better measure, since a low interest-to-income ratio indicates good payment capacity and resilience against interest-rate increases. The *debt-to-assets ratio*, that is, household debt as a percentage of total assets, is also a better risk measure, since a low debt-to-assets ratio means a high net worth-to-assets ratio and high resilience against a fall in asset values. In order to present the best possible information and to avoid giving a misleading impression, figures of these measures should be published instead. If one still insists on publishing the debt-to-income ratio, on should always also publish the better measures, the interest-to-income and debt-to-total-assets ratios.

Why does the debt-to-income ratio hardly give any information about the risks? The Riksbank often states that “[h]ousehold debt remains high, both from a historical perspective and in an international perspective. This makes the Swedish economy more vulnerable to shocks.”[1] But that household debt is high from a *historical* perspective does not imply greater vulnerability, if household repayment capacity and resilience against disturbances are good.

From and *international* perspective, one can note that that there are countries that have had higher debt-to-income ratios than Sweden without getting into any problems. There are also countries with lower debt-to-income ratios that have got into problems. Furthermore, since taxation and publicly supplied services vary across countries, it is problematic to directly compare disposable income across countries. Therefore, the household debt-to-GDP ratio is a better measure for international comparisons. According to this measure, Sweden was in 2010 close to the average of a sample of relevant countries. [2]

As is shown in this post, there is no obvious relation between the debt-to-income ratio and the magnitude of the downturn during the latest crisis. Riksbank Deputy Governor Martin Flodén has studied this issue more closely. His analysis shows that the relation may be *statistically* significant, but that it is not *economically* significant. This is because the debt-to-income ratio altogether has a miniscule effect on the expected future macroeconomic outcome. [3]

In figure 1, the grey line shows the debt-to-income ratio. (Statistics Sweden has revised the national accounts according to the new ESA 2010 regulation, which has resulted in an upwards revision of disposable income and downwards revision of the debt ratio with a couple of percentage points.) The vertical dashed line shows when Finansinspektionen (the Swedish FSA) introduced the loan-to-value cap of 85 percent for mortgages. Since then the debt-to-income ratio has been quite stable.

One reason for publishing the debt-to-income ratio is that it would give an indication of the debt service relative to disposable income. The debt-service share of disposable income is natural measure of the debt burden and the risks that the debt may imply. But then it is better to directly publish the interest-to-income ratio, the (after tax) interest payments as a percentage of disposable income. One can also show the interest-to-income ratio for different assumptions about the interest-rate level. In particular, one an assess debt sustainability by calculating the interest-to-income ratio that corresponds to a constant real debt and a constant debt-to-income ratio, respectively. (This is done in table 1 in this post.)

The interest-to-income ratio is shown in figure 2. A low interest-to-payment ratio indicates good repayment capacity and resilience against interest-rate increases. Currently the interest-to-income ratio is less than 4 percent. There is no trend towards an increasing interest-to-income ratio.

Another natural measure of any risks associated with debt is the debt relative to household assets, the debt-to-assets ratio. The red line in figure 1 shows the debt-to-assets ratio (measured along the right axis). It was less than 25 percent at the end of 2013. This means that the household net worth-to-assets ratio was larger than 75 percent. A rising debt-to-assets and thereby a falling net worth-to-assets ratio would be an important warning signal. But there is no trend towards an increasing debt-to-assets ratio.

Among these *aggregate* measures, the better ones, the interest-to-income ratio and the debt-to-assets ratio, obviously give a different impression than the debt-to-income ratio. Why not regularly publish these measures instead, or at least every time supplement the debt-to-income ratio with these better measures?

When it comes to *disaggregated* measures of the risks associated with the debt, the best information is far given in the Finansinspektionen’s Mortgage Market Report and the Commission of Inquiry of Excessive Indebtedness chaired by Anna Hedborg (unfortunately only available in Swedish).

[1] Monetary Policy Report July 2013, p. 9.

[2] Cecchetti, Mohanty och Zampolli (2011), table A2.1.

[3] According to Martin Flodén’s analysis, 1 percentage point higher debt-to-income ratio would imply a 0.02 percentage point higher increase in unemployment in a crisis (according to Flodén’s table 1, the effect is statistically significant for a sample of 26 OECD countries; according to his table 6, the effect is not statistically significant for the sample of OECD countries where housing prices fell). With such a high probability of a crises of as 10 percent per year (that is, a crisis on average every 10 years), the expected increase in unemployment is only 0.02 percentage points. With an empirically more realistic probability of 4 percent per year (that is, a crisis on average) the expected increase in unemployment is only 0.008 percentage points. A 10 percentage point increase in the debt-to-income ratio thus has a miniscule effect on the expected future unemployment. See this post and this paper for more detailed discussion and calculations.