RESEARCH IN GOVERNMENTAL AND NONPROFIT

2JACQUELINE L. RECK ET AL.research is needed in the government capital markets area to examinewhether these policies have improved the usefulness of governmental accounting information. Therefore, throughout the paper, we provide directionfor future government capital markets research.1. INTRODUCTIONIt has been over a decade since Ingram et al. (1987) provided a review of the capitalmarkets research in government accounting. Since that time additional research inthe governmental capital markets area has been conducted, and significant policydevelopments have occurred that raise new financial reporting and disclosure issuesand provide unique opportunities for additional research in this area. Consider, forexample, that institutional changes in governmental accounting and auditing havedramatically improved the quality of governmental accounting and auditing sincethe main body of this research was published.1,2 Thus, an important empiricalquestion is whether improvement in the quality of governmental financial reportinghas strengthened the association between municipal bond measures and accountingand reporting information.More recently, the Securities and Exchange Commission (SEC), throughits Municipal Securities Rule-Making Board (MSRB), significantly increaseddebt issuers’ disclosure requirements to municipal bond investors.3 As a result,there is now a strong need for additional research regarding the decisionusefulness of recent regulatory policies aimed at overcoming deficient financialreporting quality, and continuing disclosure to investors in the municipal bondmarket.Finally, the Governmental Accounting Standards Board’s (GASB) new governmental reporting model (GASB, 1999) represents a revolutionary change in thehistory of governmental accounting. The new model, when fully implemented,will provide many additional research issues, such as whether government-widefinancial information similar to that presented by for-profit organizations is moreuseful in pricing bonds than the traditional fund-based financial informationcurrently reported, and whether disclosure of infrastructure assets is useful inpricing debt.The purposes of this paper are three-fold: (1) to review the research over thepast 15 years that investigates the impact of accounting related information onthe municipal bond market; (2) to provide additional empirical research relatedto one of the more thinly studied areas – the municipal bond secondary market;and (3) to provide direction for future research.

Governmental Capital Markets Research in Accounting3In the next section we review the research published since Ingram et al. (1987)and provide direction for future research. An empirical analysis of the issueof whether government-specific financial accounting information is impoundedin secondary market prices is presented in Section 3. Concluding remarks arepresented in Section 4.2. A REVIEW OF RECENT GOVERNMENTCAPITAL MARKETS RESEARCHThis review extends Ingram et al. (1987), which focuses on the association betweenaccounting and audit information and bond market measures. The bond measuresin their study include bond ratings, new issue costs and seasoned market yields.In this section we review the subsequent governmental capital markets research,dividing the studies according to the same bond measures – bond ratings, new issuecosts and seasoned market yields. These three areas are further divided based onthe effects of financial information and reporting practices, specific accountingissues, and audit reports/practices. Table 1 provides a summary of the additionalresearch that has been conducted in each of these areas.Table 1. Government Capital Markets Research in Accounting since 1985.Financial information/reportingpracticesBond RatingsNew IssuesSeasoned IssuesLewis, Patton andGreen (1988)Wescott (1988)Wilson and Stewart(1990)Benson, Marks andRaman (1991)Fairchild and Koch(1998)Stover (1991)Feroz and Wilson(1992)Soybel (1992)Vijayakumar (1995)Copeland and Wilson(1986)Chaney (1998)Sneed and Sneed(1999)Ingram, Raman andWilson (1989)Liu and Seyyed(1991)Marquette and Wilson(1992)Ingram and Wilson (1999)Specific accountingissuesAuditreports/practicesLangsam andKreuz (1990)Raman and Wilson (1990)Raman and Wilson (1994)

4JACQUELINE L. RECK ET AL.2.1. Bond Rating ResearchAt the time of the Ingram et al. (1987) review article, more studies (e.g. Copeland &Ingram, 1982; Ingram & Copeland, 1982a; Marquette & Marquette, 1986; Raman,1981, 1982a, b, 1986; Wallace, 1981; Wescott, 1984; Wilson & Howard, 1984)had examined bond ratings than any other municipal bond market measure. Thesestudies find an association between accounting variables and bond ratings and bondrating changes. In general, however, the predictive accuracy for municipal bondratings models is about 10–20% less than that for corporate bond rating predictionmodels. The authors (Ingram et al., 1987) suggest that the observed difference isattributable to noise in government accounting data. This argument is plausiblesince the studies cited by Ingram et al. (1987) use data from the period prior tothe formation of the GASB, implementation of the Single Audit Act, and the SECdisclosure requirements.4 Poor governmental accounting and reporting practiceswere well documented during the late 1970s and early 1980s. Additionally, manyof the studies reviewed by Ingram et al. (1987) introduce measurement error sincethey rely on accounting data obtained from the Bureau of Census rather than fromactual annual reports. Icerman and Welch (1989) find that there are differences ofapproximately 7% between Census-based financial information and informationcollected directly from the municipality’s annual report.As can be seen from Table 1, relatively little accounting research has focused onbond ratings since Ingram et al. (1987). One reason is the increasing availability ofbond price data over the past 10–15 years. Since prices are continuous measuresof default risk and marketability they are preferable for research purposes overthe crude ordinal bond rating measure. Additionally, the increasing availability ofbond price data has allowed researchers to investigate bond market participantsother than bond rating agencies, such as underwriters, secondary market investors,and bond insurers.2.1.1. Financial Information/Reporting PracticesBecause prior research (e.g. Marquette & Marquette, 1986) finds that environmental variables may impact the usefulness of financial accounting information inassessing bond ratings, Wescott (1988) uses cluster analysis to divide her sampleinto three groups of cities with similar environmental characteristics. For eachof the three city groups defined in her study, Westcott conducts a probit analysisusing financial accounting variables to predict bond ratings. She finds that thesignificance of financial accounting variables varies by city group, implyingthat for different market segments accounting information may be impoundeddifferently. Classification accuracy for Wescott’s (1988) models is greater than70% for two of the three city groups. However, classification accuracy ranges

Governmental Capital Markets Research in Accounting5only from 28 to 53% when the probit model is used to predict ratings of a holdoutsample. The large number of variables, combined with small sample size maycontribute to Westcott’s inability to construct a generalizable predictive model.Recently, only one other accounting study has attempted to look at bondrating behavior. Lewis et al. (1988) use a judgment-based lens model to comparefinancial analysts’ information choices for rating a bond with predictions ofa statistical bond rating model. They provide analysts with twelve items ofinformation, nine of which are financial accounting variables and three of whichare environmental variables. Analysts are allowed to select five items5 on whichto base their rating decision. The results indicate that analysts’ predictive accuracy(39%) is comparable to the statistical models (41%) when original selections andweights are applied across time periods (Lewis et al., 1988). A limitation of theirstudy is that variable choices are imposed on the analysts. Additionally, since thereis no theory on which to base accounting variable selection, the nine accountingvariables provided to subjects may not be the most salient variables. Interestingly,the statistical models do not select any of the environmental variables, but relysolely on financial accounting variables (Lewis et al., 1988).Both of the above studies (Lewis et al., 1988; Wescott, 1988) rely on data fromthe early 1970s to the early 1980s. During this time financial information of government municipalities was less standardized and less reliable than the financialaccounting and reporting data that has become available since the inception ofGASB and the Single Audit Act. Since the financial data may have been viewedas unreliable by users, the ability to construct predictive models was seriouslyimpacted. Additionally, both studies rely on Census-based financial data, whichhas been shown (Icerman & Welch, 1989) to vary from data reported in actualfinancial statements.2.1.2. Audit Reports/PracticesLangsam and Kreuz (1990) look specifically at the impact of audit opinion on bondmeasures. They find a significant association between audit opinion and bond ratings. Bond rating is a dichotomous variable, represented as high quality bondratings (Aa and above) and lower quality ratings. The study relies on a simple correlation between audit opinion and bond rating. Because bond ratings are dividedinto two groups, the association between opinion and each level of bond ratingcannot be determined. Furthermore, it is unknown whether the association holdsif financial accounting control variables are included in the analysis.2.1.3. Summary and Areas for Future Bond Rating ResearchIn general, the few bond ratings studies conducted within the past 10–15 yearsreinforce the findings of the earlier research. The more recent studies find that

6JACQUELINE L. RECK ET AL.accounting and auditing information are significantly associated with bond ratings.The main contributions of the more recent research are the introduction of newmethodologies (cluster analysis and behavioral research) and examination of auditvariables, such as audit opinions, in addition to accounting variables.A recent study by Fitch IBCA (Bond Buyer Staff, 2000) indicates that municipalbonds may be systematically underrated. As a result, Fitch IBCA reassessed itsmunicipal bond rating process. Additionally, Moody’s Investors Service indicatedit will work to standardize its rating process (Bond Buyer Staff, 2000). The changesbeing implemented by the bond rating agencies may have implications for what andhow financial and non-financial information is incorporated into the bond ratingprocess. For example, Fitch IBCA indicated that increased emphasis will be placedon management practices (Bond Buyer Staff, 2000). Additionally, a more objective emphasis will be placed on debt and financial performance. GASB StatementNo. 34 (GASB, 1999) requires, for the first time, that government annual reportsinclude a Management Discussion and Analysis (MD&A). We expect that ratinganalysts will find the MD&A, which provides information useful in assessing management performance, (and other features of the new model such as governmentwide financial information prepared on the accrual basis) useful in evaluating thecreditworthiness of municipal bonds. As the reliability of accounting informationcontinues to improve, the predictive ability of bond ratings models shouldimprove as well.2.2. New Issues (Primary) Bond MarketSince Ingram et al. (1987), a fair amount of government capital markets researchhas been conducted in the new issues, or primary bond market. The researchincludes a variety of financial accounting variables and reporting practices, as wellas new accounting issues. No recent research has been published involving theeffects of audit reports and practices on new issue yields.2.2.1. Financial Information/Reporting PracticesIngram et al. (1987) reviewed prior accounting research that examines the importance of accounting/auditing information and the quality of financial reporting inpricing new bond issues. Those studies (Apostolou et al., 1985; Wallace, 1981;Wilson, 1983; Wilson & Howard, 1984) produce mixed results concerning theassociation between accounting variables and municipal bond yields and netinterest costs or yields. Not addressed by Ingram et al. (1987) are two studies(Benson et al., 1984, 1986) that find that variables reflecting financial reportingquality (specifically, state imposed accounting requirements and the Government

Governmental Capital Markets Research in Accounting7Finance Officers Association’s Certificate of Achievement) are significant inexplaining yields and returns under certain conditions. In the time period since1987, researchers continue to examine the association between quality of reportingand new issue measures. Researchers have also identified additional factors thatmay impact the use of accounting information – perhaps helping to explain theinconsistent results in the studies cited by Ingram et al. (1987).If financial accounting information is not considered reliable or relevant,uncertainty arises concerning the quality of management and the government’sfinancial situation. This may lead to higher (net) interest costs on new issues. Thequestion of relevant and reliable information is addressed in two research studies(Benson et al., 1991; Fairchild & Koch, 1998). The Benson et al. (1991) studyinvestigates whether voluntary disclosures by states impact the state’s net interestcost. Their results indicate that the net interest cost for high disclosure states ison average 14 basis points lower than the net interest costs for low disclosurestates. These results are supported by Fairchild and Koch (1998), who look at therelationship between municipal disclosure practices and net interest costs. Theyfind that for unrated bonds there is a differential of approximately 14 basis pointsbetween municipalities in states requiring high disclosure in official statements vs.municipalities in states with low disclosure requirements for official statements.These results are surprisingly similar given the fact that Benson et al. (1991)rely on data from October 1976 to May 1979 while the Fairchild and Koch(1998) paper utilizes data from 1980 to 1988. Furthermore, the method used toidentify high disclosure vs. low disclosure observations is somewhat different.The first study (Benson et al., 1991) relies on a six-point index based on twelveaccounting practices; whereas, the second study (Fairchild & Koch, 1998) relieson a dichotomous variable based on a late 1980s survey conducted by the NationalAssociation of State Auditors, Comptrollers, and Treasurers.Both the Benson et al. (1991) study and the Fairchild and Koch (1998) studyrely on a considerable amount of data from the pre-GASB, pre-Single Audit,and pre-SEC disclosure period. As indicated earlier, accounting information andreporting practices have greatly improved since the mid-1980s. While the questionof voluntary disclosure is still highly relevant, state regulation of disclosure mayhave less impact on new issue costs as accounting and reporting practices becomeincreasingly more standardized nationwide. Additionally, the findings of studiesin this area may be confounded if high disclosure states are found to have a higherproportion of frequent issuers, since lower average information asymmetry, andthus, lower borrowing costs would exist in these states, even without reportingregulation.Closely related to the quality of disclosure studies is a study conducted bySoybel (1992) in which she assesses the association between new issue yields

8JACQUELINE L. RECK ET AL.and cash basis accounting vs. GAAP (modified accrual) accounting in the caseof New York City. She finds that the market is not deceived by the use of cashbasis revenues and expenditures; rather the market appears to adjust to GAAPbased accounting numbers for setting new issue yields. The results of Soybel’s(1992) study indicate market efficiency, in that the market is not deceived byNew York City’s overstatement of revenues and understatement of expenditures.The generalizability of Soybel’s study is somewhat limited given the city studied.Because New York City is the largest city in the United States, it has characteristicsthat may not be relevant to other U.S. cities. Also, since New York City wasreporting on a cash basis, Soybel was required to convert to the GAAP basis. Theconversion process was necessarily crude and contains measurement error sincenot all financial accounting records were available to Soybel.Wilson and Stewart (1990) and Vijayakumar (1995) show that the level ofinformation asymmetry varies across segments of the new issue bond market.Wilson and Stewart (1990) are the first to examine the effects of informationasymmetry on the association between financial disclosure and competition amongunderwriters for new bond issues. They find that financial accounting disclosureis more important to the market competitiveness (number of bids received on thedebt offering) of infrequent debt issuers than to the competitiveness of frequentdebt issuers. Vijayakumar (1995) also finds that information asymmetry maycause “high quality” bond issuers to include a call provision, since the issuer isunable to credibly reveal its financial position to the markets.Feroz and Wilson (1992) argue that the primary bond market could be dividedinto a high information asymmetry market segment consisting of the bond issuesof smaller, less frequent issuers that are managed by local or regional underwriters,and a low information asymmetry segment managed by national underwriters.They then hypothesize that financial accounting and auditing information is morevalue-relevant to the market involved with new issues underwritten by local orregional underwriters than issues underwritten by national underwriters. Theyfind that financial accounting disclosures, type of auditor, and quality of financialmanagement are all significantly associated with net interest costs if the bondissue is managed by a local or regional underwriter. None of these variables aresignificant if the issue is managed by a national underwriter.The fact that size matters in determining net interest costs is further supportedby Reeve and Herring (1986), who find that there is an average difference of40 basis points between the net interest costs of small unrated bond issuers andlarge unrated bond issuers. Costs this large indicate that additional informationdisclosure is beneficial for smaller issuers.The effect of intermediation, discussed in more detail in the seasoned bondmarket, is also touched on in the new issues market. Stover (1991) uses path

Governmental Capital Markets Research in Accounting9analysis to investigate the relationship between bond ratings and debt relatedfinancial variables and new issue reoffering yields. He finds that debt variableshave no direct effect on reoffering yields, but instead exert an indirect effectthrough the entity’s bond rating. Again, Stover (1991) relies on 1980s data inhis tests. As the quality and amount of accounting data has improved, it is possiblethat users are more willing to evaluate and incorporate information directly intothe yields, rather than relying on evaluation by

Accounting , the only series dedicated exclusively to governmental and nonprofit accounting and reporting issues. The purpose of Research in Governmental and Nonprofit Accounting is to stimulate and report high-quality research on a wide range of governmental and nonprofit accounting top