3. Extended real interest margin model

Date



Similar to the work of Comley et al., our estimations follow an error correction model:

Formula for error correction model(1)

where algebraic formula and algebraic formula are short-run parameters. is the long-run cointegrating parameter.

The real interest margin, algebraic formula, is measured by taking the difference between 10-year Treasury bonds for Australia and the United States, adjusted for expected inflation. The real interest margin is intended to capture cross-country differences that differentiate foreign assets from domestic assets including liquidity, risk, capital flows and tax treatments.

algebraic formula is a vector of explanatory variables that includes general government net debt, the primary budget balance, the current account balance, inflation and real GDP growth.

To improve the quality of the estimates, this analysis makes a number of adjustments to the explanatory variables previously used by Comley et al. For government debt, we use general government net debt4 rather than public sector net foreign debt, as used by Comley et al. The headline budget balance variable is replaced by a primary budget balance series, which excludes net interest payments — alleviating potential causality running from higher interest rates to rising government debt-to-GDP ratios.5 An underlying measure of inflation is used to better gauge inflationary pressures, as opposed to the headline CPI measure used by Comley et al. The US 10-year government bond yields are also computed differently. For the period prior to 1997 (during which data for US Indexed Treasury bonds is unavailable), we calculate real US 10-year government bond yields based on inflation expectations.6

To extend the analysis beyond that considered by Comley et al., we include the US counterparts of the Australian variables, thereby placing greater emphasis on the extent to which external factors drive movements in the real interest margin.

A deterioration in the Australian primary budget balance is expected to cause the real interest margin to rise, while a worsening in the US primary balance is expected to cause the real interest margin to fall. Similarly, the real interest margin is expected to rise in response to an increase in the stock of Australian general government net debt, and fall when US general government net debt increases.

The real interest margin is expected to be positively related to changes in the inflation rate for Australia, with the converse holding for changes in US inflation. An average of the two underlying measures of inflation (the trimmed mean and weighted median) is used for Australia, while a trimmed mean CPI is used for the United States.

Stronger GDP growth implies that short-term interest rates need to be higher than they otherwise would be — which has potential implications for long-term interest rates. This would likely see a decrease in the spread between Australian and US government bonds.

Movements in the current account are also expected to affect the risk premium. A deterioration in Australia’s current account balance will increase the real interest margin (as the risk premium increases), with the converse holding for movements in the US current account balance.

The primary focus of this paper is on the interest rate effects of government debt from an empirical point of view. We have not investigated the degree to which government borrowing might be offset by private domestic saving or inflows of foreign saving (or both).


4 General Government debt is the total debt incurred by Commonwealth, State and Local governments combined.

5 While Comley et al. (2002) estimated the model separately using the headline and structural budget balance as flow fiscal measures, our analysis is restricted to the headline primary balance due to the difficulty of obtaining consistent measures of the structural budget balance back to the early 1990s. It would be more desirable to carry out the same estimation with a structural primary balance series to identify the effect of the cyclically-adjusted balance.

6 The measure used by Comley et al. back-casted the US 10-year government bond yield series with constant weights assigned to the bond yields, the US Federal Reserve’s federal funds rate and inflation.