Vanguard bond index fund tracking error

Tracking error is the ultimate measure of judging an index fund manager's performance. Since bond indexes can include huge numbers of illiquid bonds, replicating an index is often very costly. Vanguard, in an attempt to control costs, samples securities in their bond index funds. The sampling attempts to match the sector and risk metrics of the underlying bonds in the index. Because a sampled index fund does not hold all of the securities in the underlying index, its returns may vary somewhat from those of the index. Such performance variance is termed "sampling error." Over time, Vanguard has made a number of changes in the sampling policy used by the funds. In 2009, the funds adopted a change in benchmark tracking indices, adopting the Barclays Capital U.S Government/Credit Float Adjusted Indices.

Tracking error
Table 1. provides the long term average tracking errors for Vanguard's bond index funds. Data extends from 1995 to present for the Short, Intermediate, and Long term funds (from the first full year of operations). Data extends from 1993 to present for the Total Bond Index (first full year 1987). Admiral share class data covers the period from the 2002 inception of the share class to the present. Detailed annual tracking error data (along with additional statistical measurement of fund returns) are provided in the footnote tables.

Sampling strategies
From inception (1987 and 1995) the Vanguard bond index funds have employed sampling strategies in implementing investment policy. In addition, the funds originally employed a corporate bond substitution policy, whereby they substituted short term corporate bonds for treasury bonds in fund holdings. The funds' substitutions were limited to securities with less than 4 years remaining to maturity, and were restricted to a maximum of 15% of total assets. The rationale for substitution was that the higher interest income from corporate issues would help the funds offset the costs of implementation and would, with a moderate increase in diversifiable risk, allow the funds to more closely track their benchmark indices.

As shown in the second column of Table 2. below, the substitution policy resulted in the funds providing low tracking errors over the 1995-2001 period. In every instance, the funds provided returns that partially offset the drag of expenses. In 2002, a substantial increase in market risk and defaults hit the corporate bond market. As a result of the corporate substitution policy, the Vanguard bond index funds failed to closely track their benchmark indices. The funds held higher than benchmark weightings of telecommunication and energy company bonds, both of which were hurt by defaults and credit quality erosion. Column three, in Table 2. shows the fund's tracking errors in 2002. The short term, intermediate term, and total bond portfolios all lagged benchmark returns by over 2.00% for the year. Column four shows the tracking error of the funds over the 1995-2002 period.

Vanguard initially responded to the performance lag by reducing the allowable corporate substitution in the funds to a maximum 10% of net assets. By early 2005 the funds adopted a sampling strategy that required the funds to more tightly sample not only industry sectors, but also to tightly sample industry subsectors. In 2009, as a result of the 2008-2009 financial crisis and the expansion of Federal Reserve purchasing of mortgage backed and other debt market securities, the Vanguard bond funds adopted free float versions of the Barclay bond indices as  benchmark indices. The fifth column in Table 2. shows the tracking errors realized by the funds subsequent to the 2002 year and the changed sampling policy.

In addition, Table 2. (columns six and seven) show the tracking errors for the funds' investor shares and admiral shares over a common period (from inception of admiral shares in 2002 to present). The lower expense admiral shares can more closely track the benchmark.