The June release of the Federal Reserve's Flow of Funds Accounts for the first quarter of 1991 showed a decrease of $10.3 billion in the municipal bond holdings of households, prompting considerable discussion as to the cause and the implications for the future demand for municipal securities.
Unfortunately, the statistics upon which this discussion is based are so flawed that no firm conclusions can be drawn as to what is actually happening to municipal bond ownership.
The inaccuracies in the Federal Reserve data have been pointed out in previous article in The Bond Buyer. The total amount of municipal debt outstanding is taken from the Department of Commerce's Census of Government's which does not pick up some lease transactions and private placements.
More importantly, the census numbers do not include pre-refunded bonds as they are no longer debt liabilities on the books of state and local government.
Smaller governments are not completely surveyed, but are only sampled, introducing the possibility of error, particularly in quarter-to-quarter comparisons.
The Federal Reserve surveys the municipal bond assets of institutional investors, subtracts these holdings from the census estimate of municipal bonds outstanding, and assumes that the remainder is owned by households. Hence all the errors in the estimation process are loaded into the household sector.
The distortion introduced by this approach can be very significant. According to The Board Buyer, approximately $4.8 billion of refunding bonds were sold in the first quarter. This activity would have virtually no effect on the census estimate of municipal debt outstanding -- except to the extent that high-to-low advance refundings require somewhat larger issue amounts than the amount of bonds refunded.
Assume for a moment that all of the refundings were bought directly by individuals and 25% by property and casualty insurance companies. The quarterly Federal Reserve statistics would show no change in total municipal debt, a $1.2 billion increase in property and casualty holdings, and a $1.2 billion decrease in household holdings instead of the actual $3.6 billion increase.
On the other hand, as bonds issued in the first quarter of 1981 carried very high interest rates, many issues sold in that quarter have been advance refunded.
Pre-refunded bonds with a 10-year call were redeemed in the first quarter of 1991. In this case, the amount of bonds outstanding as measured by the Census of Governments would remain unchanged and, to the extent that bonds were called away from institutional investors, the Fed would show an increase in individual holdings instead of an actual decrease.
It is not all clear what really happened to household holdings of municipal securities in the first quarter. But if they declined by $10.3 billion as reported in the Fed statistics, it is only a coincidence.
The flaws in the Flow of Funds Accounts have broader implications. According to the preliminary results of a statistical analysis undertaken by Aaron Gurwitz of Goldman, Sachs & Co. using 1,000 randomly selected Cusip numbers, 21.2% of outstanding municipal bonds are pre-refunded.
$1.065 Trillion Outstanding?
If the census data is adjusted to reflect this finding, the total amount of municipal securities outstanding at the end of 1990 was $1.0655 trillion instead of the reported $839.6 billion. The total household holdings, as defined by the Fed, were $569.6 billion instead of the reported $343.7 billion, a 60% increase.
Actually, the Federal Reserve household estimates include municipal bond holdings of personal and unit investment trusts and closed-end investment companies. Further, individuals own municipal securities through long-term and money market mutual funds.
This increase in the estimate of total bonds outstanding is consistent tith the analysis by Bruce F. Davie, in the Feb. 4, 1991 Bond Buyer, of IRS data on tax-exempt income reported on 1988 tax returns. In fact, the amount of the upward adjustment indicates that data from estate tax returns gives an even better estimate and that tax-exempt income received by high-income taxpayers is substantially underreported on income tax returns.
It is clear the individual investor is even more important in the municipal bond market than previously thought. Individuals and their proxies hold over 70% of all outstanding municipals and -- because total pre-refunded municipals outstanding grew rapidly in the 1980s -- the increase in individual investor participation during the 1980s was even more dramatic than is reflected in the Fed statistics.
Second, though mutual funds and other forms of institutional intermediation of individual ownership have taken a much larger share of the market in the last 10 years, direct ownership still accounts for the majority of holdings by individuals and almost 40% of total municipal bonds outstanding.
Many analysts are bullish on the municipal market because the roll-off offpre-refunded debt will free funds for investment in new tax-exempt debt.
However, short-maturity triple-A rated tax-exempt debt has always been in great demand, as reflected in taxable equivalent yield comparisons with Treasury and corporate securities.
The new debt that must be issued to finance infrastructure investment is long-term and -- unless it is credit-enhanced -- of lower credit quality. It is unclear whether investors called out of pre-refunded municipals will be willing to shift their investment horizons to invest in the new money issues.
It is clear that the Federal Reserve Flow of Funds Accounts as currently compiled will tell us very little about what happens.
Mr. Ruth is writing a book for the Twentieth Century Fund on the role of the municipal bond market in financing public infrastructure.