You might read Robert Higg’s essay ‘Eighteen Problematic Propositions in the Analysis of the Growth of Government’ wherein he debunks numerous economic fallacies currently in vogue - I recommend reading the entire thing, but Proposition 11 notes the dangers of treating government as though market mechanisms apply. Feel free to ignore if this is not relevant…
Proposition 11
Putative “public demand,” especially as expressed by voting, drives the political-government system. Elected officials (and hence the bureaucracy subordinate to them) may be viewed as perfect agents of the electorate.
Adherence to this proposition characterizes the bulk of all analysis dealing with the growth of government in the West, regardless of analytical tradition or ideological leaning. (Specific citations seem unnecessary. See virtually any issue of Public Choice as well as the widely cited articles by Meltzer and Richard [1978; 1981; 1983], Peltzman [1980; 1984; 1985], Becker [1983; 1985], and Borcherding [1977; 1985]. The most recent and most extreme contribution along these lines is Wittman [1989].) This approach displays a professional deformity related to the economist’s basic tool of analysis, the theory of markets with its component theories of demand and supply. Applying their familiar tools to the analysis of politics, economists immediately look for analogues. What is the “good” being traded? Who is the “supplier” and who the “demander”? What is the “price”? The answers seem obvious. Public policy is the good; the elected legislators are the suppliers; the voters are the demanders; votes are the currency in terms of which political business is being transacted. Thus voters “buy” the desired policies by spending their votes; the legislators “sell” policies in exchange for the votes electing them to office. (See Benson and Engen [1988] for an especially straightforward application of such analogues.) Economists view consumer demand in ordinary markets as ultimately decisive for the allocation of resources; hence consumer “sovereignty,” a political metaphor imported into economics. Applying their familiar apparatus of thought to politics, economists tend to think that ultimately the political system gives the voters what they want. Therefore, if government grows, it does so because that is what the people want (Musgrave 1985, p. 306; Stiglitz 1989, p. 69). Demand creates its own supply. Voting is ultimately all that matters for determining the growth of government. As Dennis Mueller (1987, p. 142) has observed, “In the public choice literature the state often appears as simply a voting rule that transforms individual preferences into political outcomes.”
It is easy—and probably healthy—to mock this view of the political process. Joseph Schumpeter (1954, p. 429) called it “the perfect example of a nursery tale.” There are, after all, many significant differences between ordinary markets and the “political market” (Higgs 1987a, pp. 14-15). Even Benson and Engen (1988, pp. 733, 741), adherents of this model, describe their output variable as “somewhat artificial and very restrictive” and their price variable as “clearly an incomplete proxy.”
Not least of the problems is that voters rarely vote directly for or against policies. Rather, they vote for candidates for office. Winning candidates subsequently enact a multitude of policies, many of which neither the voters nor their representatives had thought about at the time of the campaign. It is not enough that voters know something about the general ideological reputation of office seekers (à la Dougan and Munger 1989); the devil is in the details. Besides, notwithstanding the elaborate theoretical and econometric attempts to show that politicians are perfect agents (Becker 1983; 1985; Peltzman 1984; 1985; Wittman 1989), we can easily demonstrate that political representatives frequently act in ways that must necessarily run counter to the dominant preference of their constituents. We see this in the U.S. Senate, for instance, every time the two senators who represent the same state split their votes—and such splitting occurs commonly (Higgs 1989d). Remarkably, and quite damningly for models that presume tight linkages between voters and their elected representatives, many of the vote-splitting senators are reelected time and again. So elections are reliable neither as an ex ante nor as an ex post check on the substantial autonomy of officeholders.
Perhaps the most important case in which legislators and other (including many nonelected) officials act independently of control by the voters concerns political action during crises. How many voters could possibly have known in the election of 1940 what the elected federal officials would do during their upcoming terms in office, which were to include, depending on the office, some or all of the years of World War II? How many voters in the election of 1972 had any idea how they wished their representatives to deal with the “energy crisis” of 1973-1974, or even that such a crisis loomed? Who anticipated that George Bush would send U.S. troops into Saudi Arabia to oppose Iraq? During crises, government officials, lacking any reliable means for discovering dominant constituent preferences, necessarily exercise more or less discretionary power. But they do act, often in dramatically important ways.
Once those actions were taken, in a world of path-dependent historical processes the course of events was changed irrevocably (Brennan and Buchanan 1985, pp. 16, 74; Higgs 1987a, pp. 30-33, 57-74). (Ratcheting growth of government spending associated with participation in global wars is confirmed statistically by Rasler and Thompson [1985], using Box-Tiao tests.) If U.S. voters actually had preferred that the nation not go to war, it was too late to rectify the legislators’ mistake in the election of 1942 – the fat was already in the fire.
Further, political actions are usually followed by carefully crafted rationalizations, excuses, and propaganda emanating from the politicians and their friends who initiated or supported the actions. (How often do politicians admit policy mistakes?) In this way political preferences, public opinion, even the dominant ideology may be altered, becoming more congruent with what has been done and thereby reversing the direction of causality usually assumed in political models. (On ideology and policy as interactive, see Higgs 1985; 1987a, pp. 67-74; 1989c, pp. 96-98.)