Politics, Bureaucracy, and Farm Credit.
Re-examining the relationships between bureaucracy and political institutions in the area of agricultural credit is worthwhile for four reasons. First, the characteristics of this policy area differ from those assumed by the predominant theory, the principal-agent model. Second, these characteristics generate a different set of relationships that can best be described as bottom-line oversight (Khademian, 1995; Behn, 1992). In such circumstances bureaucracies are given vast discretion but held to a specific performance standard that is relatively easy to measure even if the day-to-day operations of the bureaucracy are not.
Third, such policy areas encourage a different style of congressional oversight. Here is where we add "smoke alarms" to "police patrol" and "fire alarm" oversight. Fourth, agricultural credit policy is important in its own right. The food and fiber industries are economically powerful (adding $800 billion in value to the U.S. gross domestic product annually) and politically powerful (Browne, 1988; Jones, 1961; Talbot and Hadwiger, 1968; Tweeten, 1979; Ulrich, 1989). Agricultural debt totals approximately $140 billion dollars with about two-fifths of the total held by U.S. government-sponsored enterprises.
A fundamental tenet of democratic theory is that public policy should reflect a direct link between the governed and the governors; that is, public policy should be made by officials who are accountable to the people. Thus, a direct relationship is needed between unelected bureaucrats, who may exercise policy control, and the elected representatives of the people. Not surprisingly, research into the control of bureaucracies by political institutions is a growth industry (Moe, 1982; 1985; Scholz and Wei, 1986; Wood, 1988; Aberbach, 1990; Wood and Anderson, 1993; Wood and Waterman, 1991; 1994; Woolley, 1993). Much of this research uses the principal-agent model. Principal-agent models assume that bureaucrats and politicians disagree over the goals and means of public policy and, further, that political control over the bureaucrats is made more difficult because bureaucrats have access to policy-relevant information that politicians do not (Moe, 1985, 1098). In this context of goal conflict and information asymmetry, political principals are hypothesized to control bureaucratic agencies through a combination of monitoring, dispensing rewards, and meting out punishments. The research generally investigates this relationship by analyzing how bureaucratic outputs respond to changes in presidential administrations, changes in congressional partisanship or ideology, or changes brought about by periodic political events.
We suggest that agricultural credit policy, at least in regard to Congress, generally does not fit the assumptions of the principal-agent model. Instead, we see relationships that are essentially cooperative with shared goals and fewer problems generated by information asymmetry. Correspondence with the principal-agent model, we believe, should be viewed as an interval variable with congruence being a matter of degree. That is, if we place the information asymmetry and goal conflict of principal-agent relationships at one end of a continuum and relationships built around shared goals and information at the other, then agricultural credit is toward the cooperative end. There will still be some conflict (Johnson, 1992) and some information asymmetry, but they will not be the defining elements of the relationship.
Agricultural policy is primarily distributive in nature, and Congress is the lead political institution in this policy arena (Talbot and Hadwiger, 1968; Browne, 1988, 50; Meier, 1985; and Ripley and Franklin, 1991). Although presidents have some interest in agricultural policy, it is not extensive and appears to be declining (see Welborn, 1993; Meier, Wrinkle, and Polinard, 1995).(1) Within Congress most agricultural policymaking responsibility rests with the Agriculture Committees of the House and Senate. Virtually all members of these committees have family, financial, or direct constituency ties to agriculture. Over time, casework and other constituency interactions reinforce the policy knowledge of committee members. Policy knowledge is reinforced by congressional committee stability and policy stability. An examination of the membership of the House Agricultural Committee from 1950 to 1990 demonstrates this high level of committee stability. Throughout this period, 75 percent of the Democratic members of the committee in one Congress had served on the committee during the previous Congress. For Republicans, the number was 67 percent.(2)
We believe that agricultural policy in general and agricultural credit policy in particular are stable policy areas marked by goal congruence and specialized methods of oversight (Browne, 1988; Jones, 1961; Hadwiger, 1992b). We argue that members of Congress and agricultural bureaucrats share some relatively simple and easy-to-assess goals for agricultural credit (Jones, 1961; Browne, 1988; Youngberg, 1976). These shared goals and policy values lead to a high degree of consensus on means (see Johnson, 1992, who suggests that there are elements of consensus and disagreement in virtually every political-bureaucratic relationship).
Even in areas of goal agreement, political-bureaucratic relationships must still overcome information problems if adequate monitoring is to occur. In some cases, however, Congress can design agencies to permit special forms of oversight. In their analysis of the role of congressional oversight in a regulatory policy environment, McCubbins and Schwartz (1984) identify two forms of congressional oversight: police-patrol oversight and fire-alarm oversight. Congressional intervention of the police-patrol type is centralized, active, and direct (represented by formal oversight hearings), while that of the fire-alarm type is less centralized, less active, and less direct (represented by casework). In the fire-alarm type, Congress is more passive and relies upon citizens and organized interest groups to report, i.e., sound the fire alarm, about administrative conduct contrary to congressional intent. Upon hearing the fire alarms, Congress might dispatch its trucks and fire crews to the problem area(3) Lupia and McCubbins (1994) elaborate and provide support for this model.
We suggest that the typology of McCubbins and Schwartz (1984) is incomplete. We suggest adding a third category, one that we believe is more relevant for nonregulatory policy areas. In addition to police patrols and fire alarms, Congress sometimes exercises "bottom-line" oversight (Behn, 1992). To use bottom-line oversight, programs must have clear goals, support from constituents, and an accurate and obvious means of measuring how well the organization is pursuing the goals (Khademian, 1995). Government corporations such as the various electrical power authorities, the insurance underwriting agencies, and nonagricultural credit agencies are good examples.(4) As long as the bottom line is positive, Congress plays no active role. When the bottom line turns negative (a crisis), however, Congress intervenes with legislation designed to shore up the agency at risk. Continuing the McCubbins and Schwartz metaphors, we identify this bottom-line oversight as similar to smoke detectors.
The smoke-detector concept fits well with the traditional motivations of members of Congress (Mayhew, 1974; McCubbins and Schwartz, 1984). In such distributive policy areas as general agricultural policy and agricultural credit, members of Congress can use their institutional attributes to deliver goods to their constituents and claim credit (Mayhew, 1974; Lowi, 1969). Congress can establish the implementing organization and largely ignore it, as long as the bottom line is positive. For example, Congress may desire to provide credit to the farm sector in a fiscally prudent manner. Congress, in effect, says to the implementing agency, loan money but don't go bankrupt. As long as the economy cooperates and a clear "bottom line" is maintained (there are annual reports on total loans as well as the financial solvency of the system), Congress permits the bureaucracy to operate substantially free of Congressional supervision (see Khademian, 1995, for another perspective on a clear bottom line in a government corporation). If the annual reports or GAO audits indicate potential fiscal problems, however, the smoke detector activates and Congress intervenes.
This conception of smoke-detector-bottom-line oversight is distinct from police-patrol and fire-alarm oversight. Unlike police-patrol oversight, it is not periodic, does not require members of Congress to be active, and requires little investment of resources. In these ways it is similar to fire-alarm oversight, but with two important differences. First, Congress sets out the smoke detectors (periodic reports, audits, etc.); there is, therefore, greater goal congruence and greater information symmetry than exists with fire-alarm oversight. Second, Congress does not rely on interest groups and interested individuals with their policy biases to set off the alarms (Bawn, 1994). Like a smoke detector, this form of oversight is an automated system that monitors the environment for a specific set of dangers. When the bottom line is breached and the smoke detector sounds, e.g., the agency runs out of money, Congress acts. As long as the bottom line remains clear, Congress can ignore the agency, thus freeing members for other tasks.
The contrast between a bottom-line policy system and a principal-agent system is striking. Principal-agent systems are dynamic with continued interactions. Goal conflict is exacerbated by information asymmetry. Political shifts in Congress or the presidency should translate into policy change with relative dispatch (Moe, 1985; Wood and Waterman, 1991). In contrast, a bottom-line system is characterized by stability; policy is affected most by the inertia of past policy and environmental changes (e.g., changes in economic conditions). Bottom-line agencies operate in an institutional environment (e.g., independent agency status, lack of rival agencies within a department) which is conducive to congressional inattention. Political interventions (changes in law) are rare and attempted only when the agency's bottom line is threatened.
The Farm Credit System
The federal farm credit system appears to fit the bottom-line description. It had its origins in the Farm Credit Act of 1933 and eventually evolved into a loose federation of banks organized on a regional basis (Sunsbury, 1988; Hoag, 1976; Musolf, 1991). The supervising agency, the Farm Credit Administration (FCA), was first established as an independent agency, then placed under the Department of Agriculture, and finally reestablished as an independent agency in 1953 (Hoag, 1976; Sunsbury, 1988). The Farm Credit Administration resembles a government corporation. Its expenses are passed along to the banks and associations of the system (Sunsbury, 1988). It receives no budget from Congress and thus is not subject to annual appropriations hearings.
The Farm Credit Administration meets the requirements Khademian (1995) has noted for a bottom-line agency. It has a clear mandate from Congress as well as support from its member institutions (constituents), and there is a clear method of measuring how well the organization is pursuing that mandate (Khademian, 1995). As long as the agency is providing credit to the agricultural sector and there are no major problems, Congress is willing to let the system operate without intervention. When major problems occur, however, Congress intervenes with legislation designed to address the problem.
The bottom-line model suggests that the only political factor that will affect bureaucratic outputs are congressional interventions in the process, generally in the form of legislation. During periods of nonintervention, bureaucratic outputs will be determined by inertia as well as changes in the agency's economic environment. In contrast, the principal-agent model suggests that policy will change in response to changes in principals (e.g., new presidents) or changes in the policy values of politicians (e.g., changes in partisanship).
Agricultural debt has two major components that serve as the bottom line in assessing agency performance--the acquisition of debt and the institutions that possess that debt. Total debt is measured in two ways. Total debt per farm is the first indicator. This constant dollar figure was subjected to a log transformation to eliminate a somewhat skewed distribution. The second debt measure is farm debt as a percentage of total farm assets. This measure implies that debt is a concern only relative to assets and that a large and growing farm sector can safely absorb more debt than a small and shrinking one.
Who owns the agricultural debt is also of interest because ownership is related to the level of risk involved. Government programs compete with and supplement private-sector institutions in providing capital for agriculture. Banks, life insurance companies, and private individuals are the source of substantial agricultural capital. We use three measures of debt distribution. The percentage of debt held by the government measures government's total intervention into this market. The percentage of debt held by the Farm Credit Administration and the percentage held by the Farmer's Home Administration (FmHA) provide indicators of the credit risk involved. The Farmer's Home Administration is considered the lender of last resort; as its percentage of debt increases, the risk to which the federal government is exposed also increases. Another way of expressing this measure is that it indicates the degree of government subsidy in a process that could operate in the private sector.
A variety of factors can influence the level of agricultural debt and which agency holds that debt. These factors must be included in any analysis before determining if either bottom-line oversight or principal-agent oversight exists. Our review of farm credit policies suggests that the key variables are past debt, economic conditions, existing government policies, and occasionally legislative action. Past agricultural debt is clearly a major factor. Much of farm debt is carried in long-term notes; therefore, changes in public policy or changes in the economy initially affect only debt at the margin, but they can have a cumulative effect over time. To control for the slow process of policy change in agricultural debt, we include a lagged dependent variable as an independent variable in all of our equations. The lagged variables allow us to look at changes in policy outcomes while holding constant past actions.(5)
A time series with a lagged dependent variable poses special estimation problems. Autocorrelation is likely to be a problem in any time series, but normal measures of autocorrelation are biased in the presence of a lagged dependent variable (Madalla, 1992, 290). Lagging a dependent variable can also induce a moving average problem. The appropriate test is the LaGrange Multiplier which regresses the residuals of the equation on all independent variables plus multiple lags of the residuals. This test is sensitive to both autocorrelation and moving average disturbances in the error term (Madalla, 1992, 292). Each equation was tested for autocorrelation of up to five lags. When problems were found in the ordinary least squares estimates (as they were in all equations), the equations were estimated via a maximum likelihood estimator with the appropriate corrections.(6)
In building all the models, our objective was to contrast the bottom-line model with the principal-agent model. In multiple policy theories, especially principal-agent models, which allow influence in a myriad of ways with multiple actors, theory is often ambiguous. To develop as parsimonious an explanation of policy as possible, we relied on statistical criteria. The normal F-test used in model building loses a great deal of power in multiple tests. Accordingly, we minimize the Schwarz (1978) Bayesian Criterion to build models. To emphasize the contrast between the bottom-line view of political control and the principal-agent view, we examine principal-agent variables as a group (all presidential variables and congressional partisanship). The principal-agent variables, thus, get two opportunities to enter the model, first as the models are constructed (along with all other variables), and second as a joint group after the initial model is built.(7)
Independent Variables: Bottom-Line Factors
(1.) Congressional Action. The notion of bottom-line oversight suggests that Congress will intervene rarely in farm credit policy and do so with legislation designed to correct significant problems. As indicated earlier, this contrasts with the more dynamic nature of the principal-agent model. A look at the legislative history of farm credit policy offers an opportunity to test the application of the bottom-line concept versus the principal-agent model.
Our examination of the legislative history of farm credit policy identifies 1967, 1972, 1978, and 1987 as the dates of four such interventions. In 1967 a rise in the cost of borrowing funds threatened the economic viability of the farm credit system because the Farm Credit Administration was prohibited by law from charging more than six percent interest. Congress changed the authorization of the farm credit system and essentially established a private-sector corporation without interest rate caps. This law was designed to permit the system to compete with private sector firms and should be associated with an increased government share of total farm debt.
The Rural Development Act of 1972 expanded the loan powers of the agricultural debt agencies, but it was not triggered by a crisis. The objective was rural development (not agricultural credit), and the farm credit system was simply a convenient agency to use for implementation. Although this law should also be associated with an expansion of debt and an increased share of debt for the government agencies, it was not the result of bottom-line oversight.
The Farm Credit Act of 1978 was a response to a threatened farm strike by the American Agricultural Movement. The act provided for long-term financing, but the Farm Credit Administration could only provide loans at market rates. The Farmer's Home Administration, however, with its special role as lender of last resort, had a competitive advantage in servicing the high risk portion of farm credit. The act should be associated with a decrease in farm debt and eventually a lesser share of debt owned by the Farm Credit Administration but an increase for the Farmer's Home Administration.
The economic collapse of the farm credit system in the mid-1980s precipitated a crisis resulting in the Farm Credit Act of 1987 (Musolf, 1991). This act made the Farm Credit Administration an independent regulator of the farm credit system with substantial enforcement power. The objective was to put the farm credit system on a more economically sound basis. It also created another government corporation to establish a secondary market for commercial banks' farm debt. This provision essentially extended government guarantees to private loans, thus reducing the competitive advantage of the Farm Credit and Farmer's Home Administrations (Musolf, 1991, 155). The passage of the act should be associated with a reduction in farm debt and a decline in debt owned by both government agencies.
Except for the 1978 law, these variables are measured as dummy variables. The impact of the 1978 law was such that it had a nonlinear impact on many variables; so it was also measured as a counter variable in addition to a dummy intervention. The variables were coded to have an impact starting with the first crop year after passage.(8)
(2.) Economic Factors. Farm credit policy is an attempt to manipulate the economy; if the private sector could adequately supply farm credit, then there would be no need for government intervention. By establishing a set of alternative credit systems, political institutions seek to supplement the market rather than replace it. Economic factors, therefore, continue to influence the actions of the farm credit systems. Four economic variables are relevant. First, the prime rate (measured as the corporate prime interest rate for the year) will reflect the demand for funds from sources other than agriculture. As the prime rate increases, agricultural debt will also increase, reflecting the higher cost of money in the economy. Second, a substantial portion of farm debt is in land mortgages. As the cost of agricultural land increases, new farmers will have to acquire greater debt to enter the market. The relationship between land cost (measured in constant dollars per acre) and debt should be positive. Third, like other industries, farms can finance their short-term credit needs through retained earnings. The likelihood of such earnings increases as farm equity increases (measured as total farm assets minus liabilities). The relationship between farm equity and debt should be negative. Fourth, U.S. agriculture depends heavily on exports--about one-third of all farm products are exported (Browne, et al., 1992). The ability of U.S. agriculture to compete in the world market is directly linked to the value of the dollar. A strong dollar will make U.S. exports expensive and will reduce total exports and thus farm income. The relationship between the value of the dollar (measured by the Federal Reserve Index) and farm debt should be positive.
3. Other Policy Actions. Farm credit policies are not the only policies that affect farm debt. Farms operate within an elaborate web of overlapping agricultural policies. One policy is fairly evident; by holding down interest rates, as it did during the aggressive expansion in the 1970s, the farm credit system can make its loans more attractive than those of the private sector. The interest rate charged by the farm credit system should be positively related to farm debt and debt share held by government corporations. Two other agricultural policies indirectly affect farm debt--subsidies and agricultural research. Subsidies, of course, increase income to farmers and permit the reduction of debt (but see Meier, Wrinkle, and Polinard, 1995). Agricultural research affects the inputs to farming. It can increase farm debt if the research is designed to improve farm equipment or can decrease debt if it contributes to a more efficient agriculture sector but does not require capital expenditures.
Independent Variables: Principal-Agent Factors
1. Congressional Values. Congressional values are a frequent variable in principal-agent models. Perhaps the best general measure of these values is partisanship. Outside the agriculture committees, agricultural policy traditionally has divided along partisan lines (Jones, 1961; Anderson, Brady, and Bullock, 1984). Democrats generally have been more supportive of agricultural policy than Republicans. This suggests that as the percentage of congressional Democrats increases, the amount of available agricultural credit will also increase.(9)
2. Presidential Forces. Traditionally, principal-agent studies have focused on the president, operationalizing presidential impact as a series of dummy variables that represent presidential administrations or presidential appointments. This study includes dummy variables for all presidents from Kennedy through Reagan, using Eisenhower as the point of comparison. We have no directional hypotheses specified for the presidential variables. We found many references to agriculture in presidential papers (though rarely with any policy content) and a complete absence of any discussion of farm credit.
The two models designate different variables that are likely to explain policy in agricultural credit. The bottom-line model suggests that policy will be a result of past policy, economic forces, other policies that affect agriculture, and sporadic congressional legislation. The principal-agent model suggests that policy will be a result of changes in the presidency and changes in the composition of Congress.
Table 1 presents the findings for both measures of the volume of farm debt. The models reflect a modest mix of economic and policy factors. Total debt per farm is a function of prior debt, two economic factors and two political/policy factors. On the economic side, as land costs increase so does farm debt. As farm equity increases, farm debt drops. Agricultural research expenditures are associated with greater debt as would be expected if the research were focused toward capital improvements (Meier, Wrinkle, and Polinard, 1995; Peoples, et al., 1992). The striking aspect of the equation is the paucity of political factors. Of all the principal-agent variables and the bottom-line variables, only the passage of the Agricultural Credit Act of 1978 has an impact. This legislation is associated with approximately a 25 percent reduction in the farm debt, a result consistent with prior hypotheses. The principal-agent variables as a group have only borderline significance, and their modest joint impact can be dismissed as simply an artifact.
Politics, Economics, Bureaucracy, and Agricultural Debt
Logged Debt Debt to per Farm Assets Ratio Independent Variables Slopes Slopes Economic Factors Land Cost 2.158 -- (4.14) Equity -1.621 -12.33 (3.56) (11.32) Farm Interest Rate -- 1207.37 (9.64) Political/Policy Factors 1978 Act--Intercept -.216 -- (5.77) 1978 Act--Slope -- -.75 (13.15) Agricultural Research .798 13.01 (3.61) (9.28) Subsidy Payments -- -.15 Bureaucracy (3.92) Lagged Dependent Variable .567 .53 (6.62) (10.14) Schwarz Bayesian Criterion -108.21 60.53 Standard Error .050 .383 R-Square .995 .989 Lagrange Multiplier Test (5df) 2.27 1.74 Probability for LM Test .81 .88 Error Correction MA2 AR1, AR2 Joint F-Test Excluded Variables 2.38 .61 Probability of F-Test .05 .70
Note: Slope coefficient for land cost, equity, agricultural research, and subsidies are multiplied by 1,000 to aid interpretation of results. T-scores are in parentheses.
The second column of Table 1 shows the determinants of farm debt as a percentage of assets. The results are somewhat similar to those for total farm debt. Past debt is a major force as is the economic marketplace. Debt as a percentage of assets declines as farm equity increases and increases as farm interest rates rise. Neither finding is surprising. Two general policies and one congressional action also affect this measure of debt. Agricultural research again is associated with increased debt. As predicted, subsidy payments permit farmers to reduce debt. The congressional impact is again the Agricultural Credit Act of 1978. In this case it results in a change of slope in the line so that the debt-to-assets percentage declines by 0.75 of a percentage point per year. The total impact of this factor over time is to reduce the farm-debt-to-assets percentage by approximately 30 percent from its 1978 levels. With this impact working through the lagged dependent variable in a distributive lag, the impact is substantial. The principal-agent variables in this model add no explanatory power.
Table 1 reinforces the view that farm credit is better described by the bottom-line model. The significant predictors are all part of those designated as important by that model. The factors associated with principal-agent models, presidents, and partisanship are not related to the measures of farm credit policy.
Table 2 displays the models for the ownership of farm debt. The first column presents the results for the percentage of agricultural debt owned by the government. The high coefficient for the lagged dependent variable illustrates how slow to change this percentage is. One economic factor is important--the prime interest rate. As the prime rate increases, so too does the proportion of debt owned by the government. Such a relationship is expected since the government debt is likely to be cheaper than private-sector debt. Government debt based on funds from bonding authority will be cheaper whenever short-term interest rates rise relative to long-term rates. Two congressional policies affect the government-held debt percentage: the 1978 law and the 1987 Farm Credit Act. The 1978 law changes the slope of the line by 0.17 of a percentage point per year while the Farm Credit Act has a single 1.06 percentage point decrease in the percentage. With a lagged coefficient of .95 such a decrease will produce a substantial long-run change in the proportion of debt held by government agencies. After a peak of 46 percent of all debt held by the government in 1981, this figure falls to 38 percent by 1990. Again, the addition of the presidential dummies and the congressional partisanship variable to this equation have no impact.
The Distribution of Agricultural Debt
Percentage of Agricultural Debt with: Independent Variables Government Farm Credit FmHA Economics Prime Rate .327 .398 -- (5.51) (8.24) Politics/Policy 1978 Act--Intercept -- -1.096 1.862 (2.63) (6.01) 1978 Act--Slope -.172 -.105 -- (5.28) (2.29) 1987 Act--Intercept -1.057 -- -2.033 (2.48) (5.12) Bureaucracy Lagged Dependent .950 .902 .870 (44.93) (41.62) (17.46) Schwarz Bayesian Criterion 71.01 52.05 56.81 Standard Error .453 .379 .388 R2 .998 .997 .989 Lagrange Multiplier Test (5df) 3.54 2.26 2.45 Probability of LM test .62 .81 .78 Error Correction MA2,AR3 MA1 MA3, AR5 Joint F-Test 1.10 1.38 1.49 Probability of F-Test .39 .25 .21
Note: T-scores are in parentheses.
The percentages of debt held by the Farm Credit Administration (FCA) and the Farmer's Home Administration (FmHA) are presented, respectively, in columns 2 and 3 of Table 2. Since the FCA is the mainstay of government farm finance and the FmHA is essentially the high risk lender of last resort, the results of these two equations show an interesting pattern. Both, of course, are strongly affected by past practices. The FCA is positively linked to the prime rate; the FmHA is not. This pattern occurs because the FCA competes with the private sector for providing farm loans; the FmHA does not. It provides credit to farmers that the private sector does not find credit worthy. As a result, the FCA responds to the market; the FmHA does not.
The activities of the two lenders are contrasted by their responses to the two major congressional efforts to alter the system. The 1978 Agricultural Credit Act affected both the slope and the intercept of the line for the Farm Credit Administration; the immediate impact was a 1.1 percentage point reduction and the long-term impact was a reduction of 0.1 of a percentage point per year. With the .9 coefficient on the lagged dependent variable, these influences change the percentage of farm debt held by the FCA from approximately one-third in the early 1980s to less than one-fourth today. The Farmer's Home Administration responded differently. As the FCA tightened up credit after the 1978 Act, the FmHA moved in strongly. From 1978 through the mid- 1980s American agriculture suffered hard times, and a large number of farm operations flirted with bankruptcy. The FmHA increased its market share of farm debt from less than 7 percent in 1978 to a peak of 16.3 percent in 1987. Then, the 1987 Farm Credit Act, triggered by the near bankruptcy of the farm credit system, had an equally dramatic impact on reducing the market share of the Farmer's Home Administration.
Once again the principal-agent variables add little to the explanations in the model. As with the other debt measures, presidents may come and go, congressional party balances can shift, but the farm debt agencies continue to operate as they always have. Only when the political institutions intervene directly with new legislation do the agencies respond. The response is a lumbering one much like steering a huge supertanker. The farm debt agencies gradually change in the direction of the new goals created for them. The pattern fits the bottom-line model and does not correspond to the principal-agent model.
This article has examined the relationship between bureaucratic and political control over farm debt policy from 1950 to 1990. We focused on two major policy issues relating to agricultural debt policy, the acquisition of debt and the institutions that possess that debt.
Recent research contends that agriculture is "a distributive policy area marked by shared information and cooperation between Congress and the bureaucracy" (Meier, et al., 1995, 450). Following this research, we suggest that an appropriate explanation of control is found in congressional bottom-line oversight, a "smoke-detector" notion, where there is a clear mandate, support from constituents, and a clear means of evaluating how well the agency is pursuing the mandate. As long as the bottom-line established by Congress is not breached, the smoke detector remains silent and congressional intervention is minimal. If a crisis occurs, and the bottom-line turns negative, however, Congress will intervene with major legislation. The smoke-detector model suggests that, as a rule, bureaucratic outputs are influenced primarily by major congressional legislative interventions, which occur only in response to failures by the agency to adhere to the "bottom-line" mandate established by Congress.
In substantive terms, our models indicate that the bottom-line variables better explain farm policy outputs in the two areas we examine, the acquisition of farm debt and the institutions that possess that debt. Many of the influences on farm debt are the result of past practices, economics, or other agricultural policies. We were able to find only sporadic political influences on farm credit policy, consistently represented by the passage of legislation. The traditional principal-agent variables--partisanship and presidential administrations had no explanatory power.(10)
We are well aware that bottom-line oversight could be incorporated into principal-agent models by defining it as a special case of principal-agent relationships with goal consensus and little information asymmetry. This incorporation, however, does violence to the principal-agent model because the model in general requires goal conflict and information asymmetry to operate. Rather than interpreting bottom-line oversight as a special case of principal-agent models, we interpret both principal-agent models and bottom-line models as special cases of overhead democracy. Overhead democracy relies on several political institutions, each with multiple possible methods to influence the bureaucracy. Since relationships between bureaucracy and political institutions are structured by the policy environment (Lowi, 1969; Meier, 1993), we should expect that more than one type of relationship will exist between bureaucracy and other political institutions.
This article suggests that the smoke-detector (bottom-line) model is an appropriate model to apply to agricultural credit policy. We think it is also applicable to other policies that are similarly structured. Such policies are characterized by clear, agreed-upon goals and an objective bottom-line method of assessing performance. Changes in presidential administrations and congressional partisanship should have minimal impact on such agencies, and direct congressional legislative intervention will be the primary means of political influence.
(1.) For an example of presidential involvement in agricultural debt policy see the 1896 campaign between William Jennings Bryan and William McKinley. Bryan's advocacy of the silver standard was an effort to appeal to farm debtors by supporting inflation. The silver standard versus the "Cross of Gold" was probably the major electoral issue in 1896.
(2.) Tabulated by the authors from membership rosters in the Congressional Directory for each appropriate congress. For a somewhat different view of congressional dominance, see Woolley, 1993.
(3.) A significant problem with the fire-alarm type of oversight is the credibility of the reported alarm from interested groups. See Bawn (1994).
(4.) These hypotheses are complicated by how well the agencies perform. The Postal Service is structured for bottom-line management but its perceived poor performance means greater intervention. The relationships are also complicated by the mixture of tasks that agencies perform. The Federal Deposit Insurance Corporation in addition to providing insurance also regulates banks. Bank regulation is clearly not a bottom-line function but the provision of insurance per se probably is.
(5.) This specification means that all other independent variables will affect agricultural debt as a distributive lag because any impact on the dependent variable at time t will also have additional impact at time t + 1 and future time periods through the lagged dependent variable (Madalla, 1992; Wood, 1990, note 6).
(6.) We estimated all equations with the maximum likelihood estimator in SAS ARIMA. The error correction used is reported in the tables. All data and documentation necessary to replicate the analysis is available from Kenneth Meier.
(7.) We are not arguing that someone could not use the current data and build a model that looks like a principal-agent model. Clearly with the number of presidential dummy variables and a reinterpretation of the legislative variables as Congress acting in a principal role, an analyst could produce good results (though not as good as the results produced with these models). To do this, however, the analyst would have to give principal-agent variables some type of preference over other variables and would be left with a model that would be difficult to explain in substantive terms given the lack of presidential interest in agricultural credit during this time period.
(8.) The use of a lagged dependent variable changes the normal interpretation of the change in intercept and change in slope variables. As noted above, a change in intercept dummy becomes a distributive lag; the change in slope dummy results in a nonlinear change in the series.
(9.) Partisan distributions in the chamber will of course also reflect partisan distributions in the committee. If partisanship has an impact, it might be in either place, although the literature suggests that the Agriculture Committees are generally bipartisan (Jones, 1961).
(10.) We have ignored a normative issue here. Bottom-line oversight is similar to policy subsystems in that it operates at relative autonomy from the majoritarian actions of political institutions. Such relationships have been criticized as outside the control of the president or the entire Congress (e.g., Lowi, 1969; Musolf, 1993). Even if we grant these criticisms, we do not have a case of bureaucracy without democratic controls but rather a case of political institutions that are unable to behave as unitary actors. This is an interesting political question but not a question of bureaucracy and democracy per se.
Aberbach, Joel D. (1990). Keeping Watchful Eye: The Politics of Congressional Oversight. Washington, DC: The Brookings Institution.
Anderson, James E., David W. Brady, Charles S. Bullock, and Joseph Stewart (1984). Public Policy and Politics in America. Monterey, CA: Brooks-Cole.
Bawn, Kathleen (1994). "Bureaucratic Accountability for Regulatory Decisions: Comment on Lupia and McCubbins." Law and Contemporary Problems 57: 139-142.
Behn, Robert D. (1992). "Bottom-Line Government." Unpublished paper. Durham, NC: The Governors Center, Duke University.
Browne, William P. (1988). Private Interests, Public Policy and American Agriculture. Lawrence: University Press of Kansas.
--, Jerry R. Skees, Louis E. Swanson, Paul D. Thompson, and Laurien J. Unnevehr (1992). Sacred Cows and Hot Potatoes: Agrarian Myths in Agricultural Policy. Boulder, CO: Westview Press.
Edmonds, Charles P., and John H. Hand (1988). "Are Farm Loans Too Risky?" Akron Business and Economic Review 19: 45-56.
Gale, H. Frederick (1993). "Why Did the Number of Young Farm Entrants Decline?" American Journal of Agricultural Research 75: 138-146.
-- (1992). "How Economic Conditions Change the Number of U.S. Farms, 1960-88." The Journal of Agricultural Economics Research 42: 22-29.
Hadwiger, Don F. (1992a). "Technology in Fragmented Politics: The Case of Agricultural Research." Technology in Society 14: 283-297.
-- (1992b). "Who Creates Food Abundance? Agricultural Policy Decision Structures and Productivity in Developing Countries." Food Policy 17: 337-348.
Hoag, W. Gifford (1976). The Farm Credit System: A History of Financial Self-Help. Danville, IL: Interstate Printers & Publishers.
Jones, Charles O. (1961). "Representation in Congress: The Case of the House Agriculture Committee." American Political Science Review 55: 358-367.
Johnson, Cathy M. (1992). The Dynamics of Conflict. Armonk, NY: M. E. Sharpe.
Khademian, Anne M. (1995). "Reinventing a Government Corporation: Professional Priorities and a Clear Bottom Line." Public Administration Review 55(1): 17-28.
Lowi, Theodore (1969). The End of Liberalism. New York: Norton.
Lupia, Arthur, and Mathew D. McCubbins (1994). "Designing Bureaucratic Accountability." Law and Contemporary Problems 57:91-126.
Madalla, G. S. (1992). Introduction to Econometrics. 2nd ed. New York: Macmillan.
Mayhew, David (1974). Congress: The Electoral Connection. New Haven: Yale University Press.
McCubbins, Mathew D., and Thomas Schwartz (1984). "Congressional Oversight Overlooked: Police Patrols versus Fire Alarms." American Journal of Political Science 28: 165-179.
Meier, Kenneth J. (1993). Politics and the Bureaucracy. Monterry CA: Brooks/Cole.
-- (1985). Regulation: Politics, Bureaucracy, and Economics. New York: St. Martin's Press.
--, Robert D. Wrinkle, and J. L. Polinard (1995). "Politics, Bureaucracy and Agricultural Policy: An Alternative View of Political Control." American Politics Quarterly 23: 427-460.
Moe, Terry M. (1982). "Regulatory Performance and Presidential Administration." American Journal of Political Science 26: 197-224.
-- (1985). "Control and Feedback in Economic Regulation: The Case of the NLRB." American Political Science Review 79:1094-1117.
Musolf, Lloyd (1991). "Government-Sponsored Enterprises and Congress." Public Administration Review 51 (2): 131-7.
Peoples, Kenneth L., David Freshwater, Gregory D. Hanson, Paul T. Prentice, and Eric P. Thor (1992). Anatomy of an American Agricultural Credit Crisis: Farm Debt in the 1980s. Lanham, MD: Rowman and Littlefield.
Redford, Emmette S. (1969). Democracy in the Administrative State. New York: Oxford University Press.
Schwarz, Gideon (1978). "Estimating the Dimensions of a Model." The Annals of Statistics 6: 461-4.
Scholz, John T., and Feng Heng Wei (1986). "Regulatory Enforcement in a Federalist System." American Political Science Review 80:1249-1270.
Sunsbury, Ben (1988). The Fall of the Farm Credit Empire. Ames: Iowa State University Press.
Talbot, Ross B., and Don F. Hadwiger (1968). The Policy Process in American Agriculture. San Francisco: Chandler.
Tweeten, Luther (1979). Foundations of Farm Policy. Lincoln: University of Nebraska Press.
Ulrich, Hugh (1989). Losing Ground.' Agricultural Policy and the Decline of the American Farm. Chicago: Chicago Review Press.
U.S. Bureau of the Census (1992). Statistical Abstract of the United States 1992. Washington, DC: U.S. Government Printing Office.
Welborn, David M. (1993). Regulation in the White House: The Johnson Presidency. Austin: University of Texas Press.
Wood, B. Dan (1988). "Principals, Bureaucrats, and Responsiveness in Clean Air Enforcements." American Political Science Review 82: 213-234.
--, and Richard W. Waterman (1991). "The Dynamics of Political Control of the Bureaucracy." American Political Science Review 85: 801-828.
-- (1994). Bureaucracy Dynamics. Boulder, CO: Westview Press.
--, and James E. Anderson (1993). "The Politics of U.S. Antitrust Regulation." American Journal of Political Science 37:1-40
Woolley, John T. (1993). "Conflict Among Regulators and the Hypothesis of Congressional Dominance." Journal of Politics 55:92-114.
Youngberg, Garth (1976). "U.S. Agriculture in the 1970s: Policy and Politics." In James E. Anderson, ed., Economic Regulatory Policies. Lexington, MA: Lexington Books, 51-68.
Kenneth J. Meier is the Charles Puryear Professor of Liberal Arts and Professor of Political Science at Texas A&M University. A former editor of the American Journal of Political Science, he is currently working on new methods for public policy analysis, the impact of organizational structures on bureaucratic performance, and theories of policy design.
J.L. "Bubba" Polinard is Professor of Political Science and chairperson of that department at the University of Texas--Pan American. He was the primary methodologies on the current study and has continuing interests in public policies that affect minorities. His recent work has appeared in the American Review of Public Administration, the Journal of Politics, American Politics Quarterly and Social Science Quarterly.
Robert D. Wrinkle is Professor of Political Science at the University of Texas--Pan American where he teaches in the MPA program. When he is not avidly pursing his hobby of orienteering, he writes on Latino politics and public policy. He also serves as the South Texas Regional Coordinator for the Texas Educational Excellence Project.3
|Printer friendly Cite/link Email Feedback|
|Author:||Meier, Kenneth J.; Polinard, J. L.; Wrinkle, Robert D.|
|Publication:||Public Administration Review|
|Article Type:||Statistical Data Included|
|Date:||Jul 1, 1999|
|Previous Article:||Devising Administrative Reform That Works: The Example of the Reinvention Lab Program.|
|Next Article:||Bank Supervision and the Limits of Political Influence over Bureaucracy.|