The demand to regulate franchise monopoly: evidence from CATV rate deregulation in California.

Hazlett, Thomas W.

THE DEMAND TO REGULATE FRANCHISE MONOPOLY: EVIDENCE FROM CATV RATE DEREGULATION IN CALIFORNIA

The motivation to price control a franchise monopoly is examined in the context of three distinct economic views of regulatory behavior. These views are tested against data from the California cable television market, over the years 1980-85, during which a subset of monopoly firms converted from regulated to unregulated pricing for basic cable service. As the price constraints of regulation appear to be insignificant in a welfare analysis, the demand for such controls by municipalities is derived from their utility in enforcing vote-maximizing transfer schemes--a Peltzmanian political outcome with a Stiglerian economic welfare result.

I. INTRODUCTION

Since the pathbreaking inquiry by Stigler and Friedland [1962], it has been fair game for economists to investigate how the publicly stated goals of regulation square with the observed economic results. Whereas empirical research has often revealed a surprisingly low correlation, theoretical models of political behavior have been developed to explain how regulators would rationally attempt to maximize political strength (and hence personal utility) by pursuing a support-maximizing transfer scheme. How far such incentive structures for regulators go toward enhancing consumer welfare is the primary concern of public policy analysis by economists.

Indeed, the study of regulatory results has yielded new insights into the antecedent demand for regulation. When publicly announced goals are not achieved, economists are often tempted to attribute the discrepancy not to random error, but to a strategic desire to mask intended transfer schemes. Hence, results of regulation can shed light on the motives for regulation.

This paper inquires into the motivations for municipal regulation of the cable television market. Part II sets three previously developed views of regulatory behavior into a unified framework and reviews previous research on price regulation in the electric utility industry. Part III describes an efficiency test for cable rate regulation in California. In a sector dominated by municipally assigned franchise monopolies, price and output effects of regulation are likely to be most dramatic, and hence most informative. Moreover, cable's changing regulatory regime offers grist for analysis. In part IV, I examine evidence from the 1979 (partial) deregulation of CATV in California and draw inferences regarding the demand to regulate cable franchises in part V. My conclusions appear in part VI.

II. REGULATORY BEHAVIOR

At least three alternative models of regulatory behavior have been advanced in the literature. Looking at the world in the Peltzman [1976] sense, politicians (regulators) are seen as engaging in two-dimensional support-maximization. In that redistributing income (by whatever institutional method) necessarily involves losers as well as winners, some "competitive equilibrium" of transfers will be achieved by maximizing political actors (see Figure 1, panel (a)).

As a rule, regulators will transfer wealth from consumers and towards the regulated industry just up to the point at which the marginal (industry) support gained is equal to the marginal (consumer) support lost. Fortuitously, Peltzman's paradigmatic political-regulatory decision focuses upon the correlation between output price and profits of a regulated public utility (seen in the "profit hill" in panel (a) of Figure 1) and the trade-off faced specifically by the regulatory agent in gaining support from one constituency at the cost of support from another (as in the [U.sup.p] indifference curve in Figure 1). This framework dovetails with the discussion of price regulation in cable to follow.

Peltzman saw the regulator as a public policy auctioneer garnering maximum political support from both consumers and producers in establishing an optimal regulated price. On the other hand, Stigler [1971] took a narrower view of the auction's participants, postulating flatly that we would expect regulation to be designed solely in the interests of the regulated industry.(1) His underlying model was the same as Peltzman's "more general" construction; Stigler, however, assumed a political indifference curve ([U.sup.s]) wherein regulators were virtually immune to the pressures of nonindustry groups (e.g., consumers). There existed little, if any, negative slope to the regulator's indifference (isosupport) curve, in that no cost was incurred in maximizing regulated industry support. Transacting difficulties, particularly the ubiquitous free rider problem, prevented consumers from exercising any effective political demands, thus removing their interests as a constraint.

Goldberg [1976], in a modern defense of natural monopoly regulation employing long-run efficiency criteria, postulated a regulatory apparatus staffed by "benevolent agents" who loyally represent buyers' interests in long-term dealings with sellers.(2) This formulation can be adapted to Peltzman's public choice model by designating a regulator utility function possessing consumer surplus as its only argument. This would generate a set of vertical indifference curves (such as [U.sup.g]) increasing in U as price declines, with maximization subject to the constraint that the producer's supply price be met.

Plainly, [U.sup.p], [U.sup.s], and [U.sup.g] yield distinct maxima under the Peltzman, Stigler, and Goldberg assumptions concerning regulatory behavior, respectively, and offer distinct welfare implications for the regulated marketplace. By extending the regulator's price-profit equilibrium into price-quantity space (as shown in panel (b) of Figure 1), the analysis yields predictable outcomes given the various regulatory environments. The Stigler framework implies a monopoly price-quantity vector; Goldberg's a Ramsey quasi-competitive result. Peltzman's model is nondeterministic, but (if distinguishable from the other two models) falls between the monopolistic and competitive endpoints.

A similar public choice paradigm was employed by Gregg Jarrell [1978], who analyzed the trade-off of pure transfers--inefficiencies being eliminated by the assumption of first-degree price discrimination(3)--instead of working within price-profit space. In such a world, [[Pi].sub.u] + [W.sub.u] = [[Pi].sub.r] + [W.sub.r] = A (where: [Pi] = industry profits; w = consumer surplus; subscripts u and r refer to regulated and unregulated), and political support is simply a function of how much producers or consumers, respectively, gain from regulation (i.e., [S.sub.p] = [f.sub.p][[Pi].sub.t] - [[Pi].sub.u]; [S.sub.c] = [f.sub.c][[W.sub.r] - [W.sub.u]]). Given standard concavity assumptions, the regulator maximizes political support at the tangency of the political support transformation function with the isosupport line (where slope = 1) furthest out from the origin. In Figure 2, one such transformation function [T.sub.1] yields an optimum at [E.sup.*].

The three-way divergence in views can be captured in distinctly sloped political support curves. A Peltzman view of political transactions costs results in a curve such as [T.sub.1]; a Stiglerian assumption yields something close to [T.sub.2]; a Goldberg analysis, [T.sub.3]. Because the Jarrell framework is efficiency neutral, it will not offer directly testable implications for the available data. It does, however, offer a consistent way of thinking about the problem and categorizing the rival theories.

An interesting addition to these theories of regulation appears in McChesney [1987]. Relaxing the assumption that political agents perform as pure brokers, McChesney models legislators and/or regulators as maximizers often able to exploit their incumbent (public) market position to extract rents or quasi-rents from producers. This bargain generally works as a threat to regulate or tax private activity, which can be extinguished by an appropriate payment to the public agent in "votes, contributions, bribes, power, and other sources of personal gain" [1987, 105]. This activity is analytically distinct from "Stiglerian rent creation," which entails enhancing the returns of incumbent producers. Rent extraction involves rents and quasi-rents(4) not specific to regulatory protectionism and will cause a positive parametric shift in the political support function (weighted in Jarrell's model by the appropriate degree of extractable rent and constrained by the competitiveness of the market for regulators). This is shown by the curve [T.sub.1] + [f.sub.s](extra-rents) in Figure 2.

If nonproducer interest groups are allowed to claim a share of regulation-created rents, however, the McChesney observation leaves the basic two dimensional calculus unchanged; what has been added is a third interest group, regulators themselves.(5) While the primary public policy consideration involves assessing net consumer welfare effects, evidence and/or intuition as to the division of transfers from producers or consumers may be useful in studying the demand for regulation. The most useful insight of the rent extraction idea is that rents or surplus can be distributed to regulators, not only to producers and consumers, and regulators need not create new net profits in order to receive industry remuneration for their regulatory activity.

III. THE CALIFORNIA CABLE RATE DECONTROL EPISODE

In September 1979, the California legislature adopted AB699, a bill permitting cable television system operators to selectively opt out of local rate regulation (as provided for in the typical municipal franchise agreement). Systems would have to qualify for full or partial deregulation on, essentially, two counts: (1) channel capacity--only systems

with twenty or more channels were

eligible to deregulate; (2) market power--systems with more

than 70 percent penetration

(Pen = no. of basic subscribers/homes passed by cable)

were ineligible for full deregulation,

but could escape local controls to

set prices at the statewide average,

with annual increases equal to

75 percent of the CPI increase

thereafter (see California Public

Broadcasting Commission (CPBC)

[1982, A-1, A-2]).(6)

The measure was a political compromise between cable industry interests, which wanted to reduce local government authority over operators, and "public access" advocates who felt that government had a duty to promote "public interest" regulatory requirements (see Denn and Smit [1983, 1] and Margulies [1979a; 1979b; 1979c]). Hence, the law attached a price to deregulation: cable systems desiring it would pay $.50/subscriber/year to a foundation established to subsidize "local origination" programming and dedicate from one to three channels (depending upon system size) for "local community, public access, educational and government access purposes" (CPBC [1982b, A-3]).(7)

With this option to deregulate for a fee, the opportunity to observe cable systems slipping from a regime of rate regulation to rate deregulation presents itself, while a control body of rate-regulated systems exists contemporaneously.(8) If local regulators act so as to maximize consumer welfare, effective prices to consumers should be significantly lower under rate regulation and penetration (basic-cable quantity demanded for a given system size) should be higher. Conversely, a pure Stiglerian "capture" view would predict no change in price or quantity demanded. The penetration test is crucial here, as it involves an output measure which may be thought of as a proxy for welfare; as such, it is a "full package" measure which allows consumers to "vote" on both price and quality.

Panel (b) of Figure 1 identifies the simple intuition behind these price and output tests. If I assume that an unregulated monopolist will price at [P.sub.m] and produce at [Q.sub.m], then the relaxation of binding price controls will result in price increases of ([P.sub.m] - [P.sub.p]) or ([P.sub.m] - [P.sub.g]), and quantity demanded will decrease by ([q.sub.p] - [q.sub.m]) or ([q.sub.g] - [q.sub.m]), depending on whether the world is best described by Peltzman or Goldberg-type assumptions, respectively. Two problems could complicate this analysis, however.

First, if demand were perfectly inelastic between [P.sub.g] and [P.sub.m], identification for the output test would be lost. I exclude this possibility, first by resorting to empirical evidence that the demand for basic cable service is not perfectly inelastic in the relevant range; the national average penetration is 64 percent (i.e., far less than 100 percent) and is highly variable over time and cross-sectionally. Secondly, an unconstrained monopolist (i.e., the cable operator after deregulation) will optimally price where demand is elastic. Indeed, using the Lerner Index as a guide, and taking the national average variable cost in CATV operations (= 1 - operating income as a percentage of revenue = .72) from Kagan [1985a, 95], reveals that [Mathematical Expression Omitted]. Hence, I assume that a significant quantity response will be prompted by a price fly-up to unconstrained levels, if controls were serving as binding constraints.

The second problem concerns the qualitative changes a cable firm would make in response to controls (and would hence "unmake" with the removal of controls). It is well known that a typical supply reaction to price controls is to lower product quality, so as to effectively raise price per output unit surreptitiously above the regulated level. Hence, a higher deregulated (nominal) price may not signal a higher effective price per quality unit. More revealing information is contained in the output changes. If both price and quality increase with decontrol, but quality increases by an equal amount or more (as measured by demand prices for the underlying characteristics), this will lead to no change, or an increase, in quantity demanded. So long as all consumers value the quality enhancement similarly, consumer surplus changes will correlate precisely with quantity changes.

Can a simple output statistic then adequately serve as a proxy for consumer welfare?(9) The specific question to be quantitatively investigated herein--the impact of price decontrol--allows me to split the welfare question into halves. Producer surplus (rents and quasi-rents to a franchised monopoly supplier) will be predictably affected in a nonnegative direction; unconstrained maximization cannot be less profitable than constrained maximization. So the possible divergence between the direction of change in the output statistic and that of total welfare would occur where consumer surplus losses more than offset producer gains.

If price and quantity react to decontrol in opposite fashion, there is an unambiguous welfare effect: when price rises and quantity falls in response to decontrol, ceteris paribus, some degree of welfare enhancement appears to be associated with regulation.(10) (Note that binding price regulation may cause quality depreciation, however, as in Leffler [1982], thereby violating the ceteris paribus condition.) The interesting case occurs where nominal price rises, but quantity does not fall. This result can obtain if product quality--at least for marginal units--rises in value so as to fully offset the nominal price increase.

As seen in Figure 3, if demand shifts up parametrically, from AB [bar] to CD [bar], then consumer surplus is unaffected by the combination of a nominal price increase and zero net change in output as EBA = FDC. (And, obviously, any increase in quantity would raise consumers' surplus.) Hence, welfare and quantity are positively correlated under the assumption that demand curve shifts (due to quality changes) take place in parallel fashion. This result obtains by assuming that incremental quality improvements are equally valued by consumers. As this is less restrictive than the commonly employed assumption of consumer homogeneity, I feel it is not an overly restrictive qualification to make.(11)

There are compelling reasons for anticipating that quality enhancement will increase after deregulation, with the demand price for cable rising accordingly. Keith Leffler [1982] models quality (q) and quantity (x) as distinct product characteristics. If q and x are net complements, welfare effects from increases (decreases) are reinforcing; if they are substitutes, however, it is possible that changes may be offsetting. For instance, rent controls (to cite an example explicitly considered by Leffler) can reduce the quantity of rental units supplied at a given quality, but if q and x are substitutes, will result in some increase of lower-quality rental space which could more than offset the decline in higher-quality units ("imposition of a maximum rental price per square foot can then result in an increased square footage of apartment space [of lower quality]" [1982, 962]). If such an effect were to obtain in the cable market, it would imply that binding price controls were lowering mean cable quality. As controls are relaxed, just the reverse process would increase quality.

A separate set of issues involves a product tied to basic cable service: premium channels. With certain restrictive assumptions, testable hypotheses could be deduced concerning the reaction of premium prices (decontrolled throughout the entire sample period) to deregulation of basic prices. Such implications are more tenuous, however, and of secondary importance for our purposes here. The statistical effects will, however, be reported in the tables.

I now consider the following hypotheses. [H.sub.0]: franchise regulators set monopolistic prices, implies: (1) [Mathematical Expression Omitted] [right arrow] the effective price of basic

service, for any given

system i, is unaffected by

regulation; and (2) [Mathematical Expression Omitted] [right arrow] the basic penetration

rate, for any given

system i, is unaffected

by regulation. [H.sub.1]: franchise regulators constrain monopolistic prices(12) implies: (1) [Mathematical Expression Omitted] [right arrow] the effective price of basic

service, for any given

system i, is greater when

unregulated; and (2) [Mathematical Expression Omitted] [right arrow] the basic penetration

rate, for any given

system i, is less when

unregulated.

These tests mirror an approach to evaluating regulatory effectiveness employed elsewhere. In Stigler and Friedland [1962], statewide mean prices and per capita outputs were regressed against state-regulated and non-state-regulated electric utilities in 1922, when a cross-sectional sample involving significant numbers of both regulated and "unregulated" states existed. While the paper found insignificant differences (but see DeAlessi [1974, 12-14] for a reinterpretation of the data), the test was not structured to strictly reveal a regulation effect, in that all electric utilities were price controlled, but a state commission regulation effect. Finding that municipalities (the default regulatory institution) set rates at about the same level as did state public utility commissions does not get directly at the effects of regulation per se. When Jarrell [1978] also compared municipally regulated to state-regulated electricity rates, he attempted to plug this hole by arguing that contemporary writers (circa 1915) often remarked upon the ineffectiveness of local franchising policies. Local regulation was only nominal, leaving virtually an open-entry, market-priced environment.

Moore [1970] chose to abandon the regulation/no regulation specification, inferring optimal monopoly prices (from estimated marginal cost and demand data on existing systems) as a proxy for an unregulated price level. Comparing such predicted prices to existing regulated prices yielded the result that the constraining effects of electric utility price regulation were visible but small--between 1.6 percent and 5.7 percent (one of four specifications yielded a negative regulatory impact; i.e., unrestricted monopoly would have been lower in the short run than the regulated price, which I exclude as implausible). While this approach did not account for the dynamic or institutional costs of regulation (such as over-capitalization, rent seeking, etc.), it provided an estimate of the magnitude of the current period savings from price regulation. As DeAlessi concludes from this (and other) evidence: "The regulation of privately-owned firms seems to yield, among other things, a slightly lower structure of rates which is more favorable to the larger users . . ." [1974, 36].

IV. THE EVIDENCE

The evidence used in this study is annual data on basic and premium cable prices and basic cable penetration for all California pay cable systems (obtained from Kagan [1978-1985]). These data, when matched with a listing of deregulated systems (obtained from the Foundation for Community Service Cable Television in San Francisco), yield a comparison of mean annual price differences for first-year deregulated systems versus the control group consisting of all non-deregulated systems.(13) Without inquiring as to differences between systems, these data reveal how particular systems change their pricing structure as they go forward in time, with some subset popping out of the regulated and into the deregulated column each year.

The key evidence consists of mean price and penetration changes of cable systems in their first year of deregulation; annual price increases of deregulated firms should not differ from others, except insofar as the former category captures the adjustment to a new regime of deregulation. This regime switch would most likely register on prices as soon as the constraint is removed. (As all data are year-end, the average first-year deregulated system would have several months to adjust.) In that firms will pay a nontrivial surcharge under AB699 to opt out of regulation, it would appear perverse if firms deregulated themselves not to raise rates. This, incidentally, is why mean price increases of first-year deregulated systems are expected to form an upper bound on the effect of rate regulation: the charge for deregulation produces selection bias. Firms which plan to take advantage of deregulation will likely be those planning to absorb the costs of such via immediate price increases.

The attractive property of observing mean price and penetration changes by regulatory policy is that it abstracts from cost, quality, or demand differences between markets. Assuming cost and demand conditions to remain constant as systems go from one year to the next is uncontroversial; the macro climate changes, as does the cable market generally, but this is what the nonderegulated system group is employed to control for.

The evidence presented in Table I (part of which is shown graphically in Figures 4 and 5) enables me to conduct a series of nonparametric tests(14) of the null hypotheses. For simplicity, I will consider the change in the price (or penetration) of first-year deregulated systems minus the price (or penetration) change in nonderegulated systems for the same year, weighted by system size (i.e., number of basic subscribers). I define the weighted average basic price increase for regulated or first-year deregulated systems in year j as (1) [Mathematical Expression Omitted] where (2) [w.sub.ij] = [no. of basic subscribers of

system i in year j] [Mathematical Expression Omitted]

 d = first year deregulated systems
 r = all systems not deregulated
 under AB699

[n.sub.r,d] = number of regulated or first

 year deregulated systems in
 given year(s)
 p = price of basic cable service

To summarize the effect across all six years in the sample, I define the mean difference in deregulated prices versus nonderegulated prices as (3) [Mathematical Expression Omitted] where (4) [Mathematical Expression Omitted] Using these measures, I take the hypotheses in order. (1) Regulated Basic Price Change vs.

First-Year Deregulated Basic Price

Change. Mean basic nominal price

increases were larger for the

deregulated systems; the data reveal a

mean annual price increase (D [bar]) for

first-year deregulated systems 10.22

higher than for same-year control

group price increases, weighting each

system by number of subscribers.

Reducing the sample to just fully

deregulated systems(15) (forty-eight of

the sixty-seven) produces a slightly

higher D [bar] = 10.87 percent. Assuming a

normal distribution, either difference is

significantly greater than zero at the

99 percent confidence level. (2) Regulated System Penetration Change

vs. Deregulated System Penetration

Change. Weighting each system by

subscriber base and summing

same-year differences (as in [3]),

penetration grew slightly more (by

0.57 percent) in all first-year

deregulated systems and by 0.18 in

fully deregulated systems. Both

differences are statistically

insignificant at accepted confidence

levels.

These results need to be squared, as the data reveal a price increase without an accompanying quantity reduction. For theoretical and empirical reasons given above, I do not suspect that this stems from a perfectly inelastic demand curve. Instead, the evidence implies a shift outwards in the demand for cable services offered by deregulated suppliers. This could only plausibly come from an increase in some dimension(s) of product quality, such as increased channel offerings on basic, service enhanced marketing effort, etc. At least at the margin of the demand curve (i.e., where [Mathematical Expression Omitted], an upwards shift of 10 percent in consumers' demand prices would be sufficient to keep output constant at nominal prices some 10 percent higher.(16) Adding a small number of channels, or collapsing "expanded basic" into basic packages (as is commonly done during deregulation) would account for such changes. While the Kagan data do not list the number of channels included in basic service, 1982 state data shows that in the first two years of AB699, decontrolled systems added 43 percent more channel capacity while regulated systems increased 27 percent (CPBC [1982a, 72]).

As price controls are relaxed, systems obtain one more (unconstrained) instrument with which to optimize. This promotes some quality enhancement, as a substitution is made away from lower-priced, lower-quality service.(17) The lack of a penetration response is key evidence tempering the conclusion to be drawn from the apparent effectiveness of a statistically significant 10-11 percent price response to deregulation. If real prices were to be raised by such an amount, penetration reductions of observable levels would predictably accompany them. This conclusion is strengthened by the direction of selection bias. In that the most eager to deregulate are likely to be those systems whose prices are most constrained, the level of regulatory price constraint is likely to be overstated here. And it is additionally revealing that apparently less than one-half of the systems eligible for regulation selected the option over the five-year period.

It is necessary to consider three alternative explanations for the low level of effective price and output response to the occurrence of system deregulation. First, the difference in penetration rate increases may be biased if the sub-groups differ systematically with respect to absolute penetration levels. Because it takes only half the added output for a system with 40 percent penetration to match the annual percentage increase of a system with 80 percent penetration, and because the lower penetration system has so much unsold market to tap (the 80 percent penetration system can increase output no more than 25 percent), growth rates should be higher for the group with lower average penetration to begin with, all else equal. Here, any such bias works against the deregulated penetration increases, however, as deregulated systems tend to have higher than average penetration rates, as seen in Table II. Even when deleting systems with greater than 70 percent penetration, the fully deregulated systems generally average (weighted or unweighted) above the statewide basic penetration mean over the 1980-85 period.

Secondly, the mere existence of the option to deregulate, i.e, the simple presence of AB699 as law, may force local regulatory authorities to loosen their rate controls. Faced with the legal possibility of total decontrol if the cable operator became sufficiently disgruntled with the regulated level of prices, the regulator would be forced to ease the legal constraint so as to retain any influence over the incumbent whatever.

This argument implies that California cable prices would move towards unregulated levels not as systems popped into the deregulated column, but as the law enabling them to do so took effect. This prediction may be empirically examined by comparing California rates statewide to cable rates nationally in the years following the passage of California's unique cable deregulation act. Mean California basic rates did increase relative to U.S. rates in the 1979-1981 period (see Table III). In 1979, the year AB699 was passed, California basic prices were 3 percent above the national average; by 1981 they were 10.2 percent higher. The conclusion that AB699 was responsible for the increase, however, is contradicted by the parallel movement of pay rates. In 1978, California premium prices were 1.5 percent below the national average; by 1980 they had increased to 13.9 percent above the national mean. An effective California deregulation of basic prices unaccompanied by a fall in penetration levels (see Table I) would not have raised California pay rates (vs. the U.S. mean). Apparently a demand shift (outwards) for cable services in California, relative to the national marketplace, occurred during this period. This is consistent with the accompanying increase in basic penetration rates; from 31 December 1978 to 31 December 1981, California penetration increased from 49.4 percent to 53.6 percent (Kagan [1978; 1981]).

A third consideration is that deregulated firms may not immediately opt to raise prices. A firm could elect to defer rate increases for strategic reasons; diplomacy in franchise renewal matters providing the prime motivation (see Olmstead and Rhode [1985]). A test for a lagged deregulation effect can be fashioned by observing price increases by deregulated systems occurring in the first two years of deregulation versus the control group.(18) These results are summarized in Table IV. They are similar to the first-year deregulation results, given that the 1985 decontrol sample is lost. The first two years' difference in deregulated systems' price increases is a modest (but significant) 5.58 percent above the associated control groups' for basic service. Penetration increases for deregulated systems, however, average 2.12 percent above the regulated systems. Here regulation appears even less constraining in its price and output consequences.

The conclusion that deregulation had a negligible effect on cable rates is reinforced by the observations of informed participants in the California deregulation episode. Monroe Price, a UCLA law professor who in 1979 "help(ed) write a measure [AB699] which [Governor Jerry Brown] eventually signed...reflected candidly that the law he shaped really 'didn't do very much and it didn't cost very much'" (in Soble [1982, I-3]). Moreover, the legislature's own study of the deregulation episode, assigned to the California Public Broadcasting Commission in 1982, found

...rate regulation "innocuous," that

rates are lower in deregulated

communities than in regulated

communities; and that the tradeoff...between

rates and community service was in

practice not effective (in Denn and

Smit [1983, 2]).

V. THE DEMAND FOR CABLE PRICE REGULATION

The evidence shows an effect on nominal price levels from rate regulation which does not translate into increased output. Hence, I conclude that effective welfare enhancement is not the result of such local price controls. Given that Moore [1970] was to find only small constraints from state regulation of electricity rates, and that local cable regulators have, a priori, less opportunity effectively to suppress prices, this should not be a surprising result.(19)

These data tend to support the Stiglerian model of the consumer welfare effects of regulatory activity, if not its behavioral assumptions. Regulation, at least in this instance, does not limit prices below profit-maximizing levels, given a particular service package. This coincides with the monopoly output result in Stigler's model, but the rents thereby created are not exclusively claimed by producer interests. Whereas Peltzman's theoretical formulation is more general, it explicitly confronts the political actor with a two-dimensional world: consumers versus supplier(s). In a complex, competitive political environment, however, gains need not be distributed solely to sellers and buyers of the regulated product. Regulation of product price is not the only policy instrument available to regulators, and consumer surplus associated with constraining product price not the only argument in the political objective function.

A pure Stiglerian capture model would imply no change in the output price or quantity to be associated with deregulation of systems: captive regulators do not shackle their masters, and so deregulated firms are not unshackled. Conversely, the Goldbergian public interest thesis predicts output restriction resultant from effective price increases to accompany deregulation. The significant nominal price change coupled with an insignificant output change (and in the wrong direction) found here suggests that neither view of regulation in this market is correct. But neither does the Peltzman compromise solution appear to obtain, in that consumers do not respond favorably (with increased quantity demanded) to rate-regulated services.

What has made the analysis confused is the ready transformation of political trade-offs into price-product space. Evidence of some--but less than full--capture does not necessarily imply a trade-off between consumers and the regulated industry. While Stigler's monopoly solution may be the result of regulation for suppliers, the rents thus generated may also be split amongst various nonconsumer constituencies.(20) With positive transactions costs, the price control institution will itself induce a lag on price adjustments without causing Pareto improvements. Instead, product quality improvements will be delayed by regulated producers who are unregulated in the quality dimension. The outcome is a monopoly welfare result with lower product quality, accompanied by a rent-sharing arrangement in which transfers are collected by the monopolist but distributed according to effective political demands.

In the CATV marketplace, the franchise auction has been modelled as a transactions-cost-minimizing institution to facilitate support-maximizing cross subsidies (Posner [1971], Hazlett [1989]). Monopoly rents created for a successful bidder may be diverted to organized groups outside of the general class of consumers; winning franchises do not bid Ramsey optimal prices, as in the hypothetical experiment conducted in Demsetz [1968], but grants, stock, and subsidies-in-kind to influential pressure groups as in Hazlett [1986b], Beutel [1989], and NTIA [1988](21). This ability to capture support from diverse interests via franchise competitions allows the regulator to put many constituencies ahead of the "consuming public" other than simply the chosen monopolist itself. This multidimensionality of rent seeking forms the essential intuition of Becker [1983].

The question arises: Why price controls? If rent maximization is the aim of regulation, franchise monopoly (sans price controls) should be sufficient. The trick is that while the Stiglerian welfare result of monopoly pricing (or rent maximization) may obtain, the motivation may not be to extract rents for the producer, as anticipated. Even given a monopoly output result, price regulation may be in the interest of nonindustry groups, including incumbent office holders. Ironically, in adopting the Peltzmanian intuition of a politician/regulator maximizing support by balancing interests, the Stiglerian cartel (or monopoly) welfare outcome is achieved when nonindustry interest groups dominate general consumer interests.(22) In allowing the regulator to effectively internalize such demands by taking payments from such organized constituencies, the McChesney model of rent extraction does just this.

If the regulator's choice of policy instruments (for instance, whether or not to impose effective price regulation on municipal franchisees) is modeled as a two-step decision, the trade-off faced by regulators in either the price-quantity space (as in Figure 1) or the producer-consumer transfer space (as in Figure 2), is seen as stage one. At this level, regulators determine output price and entry rights and, therefore, gross economic rents. In stage two, regulators consider the distribution of these rents--a trade-off in another space altogether, as in Figure 6.(23)

Assume that for some fixed level of monopoly profit the regulator will face a distributional trade-off involving competing nonconsumer interest groups. In this model, [S.sub.p] represents producer interests (e.g., cable franchisees), while [S.sub.np] denotes nonproducer, nonconsumer interests (e.g., public access advocates). Using the political support transformation function (T) in Jarrell [1978], and having determined the gross level of rents at the previous stage, the relevant trade-off now shifts to how much support from organized nonproducer, nonconsumer interests can be won in exchange for transfers, at the cost of producer support. This trade-off begins at the origin, because if no transfers are made, the producers (franchisees) collect the gross rents in whole and the regulator receives no support above that received in stage one of the regulatory game. It extends, concavely, into the lower right quadrant because all such transfers to nonproducer interests are a reduction in the net rents available to producers. The function would look something like T' and political support would hence be maximized at point E' on [Isosupport.sub.1].

Price controls can be seen as an institutional device lowering the political cost of such transfers from producer to nonproducer constituencies; hence, when they are lifted by outside policymakers, the political support transformation function recedes from T [prime] to T [double prime]. (An exogenous abolition of price controls would collapse the T-curve to a single point at the origin, were such a shock to come between discrete contract negotiations and were no other transfer enforcement mechanisms available.) This steeper slope reflects the fact that arranging the transfer of rents from producers to organized nonproducers is now done less reliably, and any given level of announced transfers will generate less nonproducer support. This, in turn, prompts a fall in political support from E [prime] to E [double prime], hurting both regulators and organized nonproducer interests, while helping producers. In effect, this two-stage analysis combines the Stigler and McChesney approaches by modeling regulators as first creating and then extracting rents.(24)

Some interesting dynamics of this process were demonstrated in the California CATV rate deregulation experience. First, the cable industry sought relief at the state level from local rate regulation and offered to pay--in the form of subsidies to local origination/public access programming--to escape from price controls.(25) Revealingly, the Public Broadcasting Commission was then appointed to study the deregulation measure's consequences by the Legislature, indicating precisely which constituency was politically vested in the outcome.

The clearest signal was then sent when the public broadcasting authorities, finding only "innocuous" effect from deregulation, chose not to conclude that regulation was a wasted effort. Instead, they lamented the trouble to be had in forcing cable system compliance with promised subsidy arrangements and suggested that the 50 cents per-subscriber--per-annum decontrol fee be levied on all systems, and/or that more deregulation be encouraged by relaxing the less-than- 70 percent penetration and twenty-channel requirements (CPBC [1982a, 106]). Moreover, their report concluded that, without rate controls, local governments would need additional tools to gain compliance in the provision of noneconomic services, such as "financial sanctions...applied to rate-deregulation systems as enforcement mechanisms" CPBC [1982a, 107]).

Price controls can be employed as convenient sanctioning devices which police concessions to interest groups, even where they do not suppress quality-adjusted prices. Here, the regulated industry was able to out-bid the opposing interest group--the municipalities themselves--at the state legislative level, obtaining freedom from rent extractions via rate decontrol. The local regulators opposed such decontrol, not out of concern over rising prices to consumers, but due to decreased ability to transfer rents to favored constituencies.

This conclusion is even more compelling in that the political interests involved in the AB699 legislation anticipated just such effects. In a time when cable systems were expanding from twelve or twenty channels to thirty-five or fifty-four channels, operators sought decontrol in the state legislature because elsewise "rate increases needed to finance such expansion turn into a political or bureaucratic nightmare in the local government" (Margulies [1979c]). Rather than argue that such freedom would result in an inefficient monopoly price level, the industry's political opposition--the League of California Cities--" argue that periodic rate hearings are the only practical method a local government has to insure that cable companies live up to the promises they made in the original franchise" (Margulies [1979a, V12]). The municipal regulators doubted that AB699's provision of $500 fines for deregulated systems which failed to provide promised public access subsidies was sufficient: "'It (rate regulation) is the biggest stick we have, and if you lose it, it makes it much more difficult to force compliance with the terms of the franchise,' said Bill Keiser, general legislative counsel to the League of California Cities" (Margulies [1979a, V12]).

Seen in this light, price controls are important institutional tools for regulators even if they generally end up allowing market prices to prevail. Rate regulation allows franchising authorities to remain "in the loop," exercising some level of control over monopoly rents which they have created and assigned. The ability to deny future rate requests gives local authorities leverage over franchise monopolists which they hope never to have to use. As such, price controls emerge as low-cost enforcement mechanisms(26) allowing transfers to nonindustry rent seekers to be reliably achieved.

VI. CONCLUSION

Whereas previous studies of the effects of price controls on franchise monopolies have compared state-regulated to locally regulated rates (as in Stigler and Friedland [1962], Jarrell [1978]), or inferred profit-maximizing monopoly prices from existing regulated systems (as in Moore [1970]), this paper has observed market price and output reactions when firms go from a regime of regulation to one of laissez-faire, all within the context of franchise monopoly. The results, however, mesh with the empirical findings of these previous studies and allow an extension of the theory of economic regulation. Due to increasingly sophisticated modeling of the regulatory process, important implications can be gleaned from the selection of institutional forms by political agents, even in instances where direct allocational results are absent.

When viewing regulation in a two-dimensional regulatory space juxtaposing consumer and producer interests, the analyst must be careful in translating observed results into welfare space. Pro-producer results may well be observed which do not map directly into a fully captured market, and vice versa. (Although a proconsumer result in price-profit space would generally imply a public interest regulatory apparatus, as dispersed consumers are ineffective brokers for rent-sharing arrangements.) In general, a two-step inquiry should ask, first, about the welfare effects created via regulation and, second, about the distribution of any regulation-created rents, in order to fully explain observed economic evidence.

The most general result of this inquiry is to suggest analysis of regulatory behavior beyond the immediate consumer effects of such policy tools as price regulation. The evidence that monopolistic output restriction was not associated with the abolition of price controls prompted a look beyond simple price effects so as to discover the purpose of rate regulation. It is apparent, in hindsight, that such tools can be important in dimensions other than price-quantity space; here price controls appear as transfer enforcement instruments. Where analysis of economic regulation, therefore, focuses exclusively upon how price moves between [P.sub.m] and [P.sub.c] on a given demand curve, it may miss the fundamental importance of the institution, and the rent-seeking competition to affect that institution, altogether. [Tabular Data 1 to 4 Omitted] [Figures 1 to 6 Omitted]

(1)"[A]s a rule, regulation is acquired by the industry and is designed and operated primarily for its benefit" (Stigler [1971, 3]). (2)Goldberg actually billed his contribution not as a defense of regulation, but as "the case against the case against regulation" ([1976, 444]). The thrust of his "plausible efficiency rationale" for regulation, however, is to evaluate such institutions as proconsumer responses to dynamic contracting problems. Goldberg specifically rules out agency or other transacting problems in constraining regulators (see Hazlett [1986a]), yielding, in fact, a political indifference curve [U.sup.8]. (3)This differs from Becker [1983], who explicitly employs such inefficiencies to reach predictions regarding the welfare losses associated with regulation. (4)Extracting quasi-rents involves long-run credibility problems; specific investments will be elastic with respect to the propensity of political extraction (as seen vividly in third world economies). At any given equilibrium level of investment, however, the opportunity for some amount of political extraction will exist (if causing higher per unit economic and political costs elsewhere). (5)Becker [1983] creates an n-dimensional analysis where various constituencies can compete for rents in addition to consumers and the regulatees. So McChesney's opportunistic regulators model can be seen as an important special case of this broader formulation. (6)It is unclear as to how vigorously the partial state-imposed price restraints were enforced. Kathleen Schuler, executive director of the Foundation for Community Service Cable Television in San Francisco, a private organization given official monitoring responsibilities over the deregulation by the Legislature, stated that "for all practical purposes, they [AB699 cable systems] were all fully deregulated" (phone conversation with the author, 7 November 1987). The Foundation's official list of deregulated systems makes no distinction between full and partial deregulations. It also appears that some of the deregulated systems with above 70 percent penetration had rates above the statewide mean, an apparent violation of AB699. Note also that the channel capacity rule was essentially nonbinding, as an operator with fewer than twenty channels of capacity could elect to upgrade its system, as was exceedingly common during this era due to the emergence of a multitude of profitable satellite services. (7)The legislation was originally crafted by cable industry lawyers and reflected an unambiguously pro-industry tilt until Governor Jerry Brown vetoed the measure in 1978. The following year, a legal aide to the governor who had been involved with public access television efforts inserted the trade-off provisions, including the $.50/ subscriber payment and the modified deregulation in high penetration systems (see Margulies [1979a; 1979b; 1979c]). (8)The franchise monopoly status of the cable systems was unchanged throughout; "deregulation" did not open local markets to free entry. This unregulated but legally protected monopoly status, since 1987 the national norm as well, is discussed in Fogarty and Spielholz [1985], Zupan [1987], and Hazlett [1986b; 1989]. An older but interesting historical account is given by Besen and Crandall [1981]. (9)In other words, will [partial derivative]CW/[partial derivative]Q > 0, where CW = consumer welfare = producers' surplus + consumers' surplus? (10)To determine the net welfare effect of regulation would require some estimation of the costs associated with the regulatory institutions, possible dynamic or rent-seeking effects, etc. Also, I theoretically rule out a situation where price falls and quantity expands after decontrol, on the grounds that it would be to no interest group's advantage to pursue price regulation which created output restriction in excess of what is available to an unregulated monopolist. (11)A deregulated firm could invest intensely in services not of value to inframarginal customers (perhaps, advertising), so as to profitably raise the demand curve nonparametrically. If such expenditures by a deregulated firm resulted in demand shifting from AB [bar] to AD [bar] then consumer surplus would increase by AGD, while the firm would be willing to expend a sum greater than this were [[Pi].sub.AD] > [[Pi].sub.AB]. The costs of private rent-seeking behavior would then exceed the comsumer surplus benefits (possible under the assumption of no price discrimination), total welfare would decline as output stayed constant (or even rose somewhat). This is the situation our parametric shift assumption rules out. As the regulation of cable television (or other industries) has not been advanced as an institution designed to economize on quality variables which are valued heterogeneously by consumers and are therefore excessive with respeect to total welfare calculations, this does not appear controversial. (12)While the null hypothesis is implied by Stigler's political model, this alternative hypothesis collapses both the Peltzman and Goldberg regulatory solutions into one, as an empirical first approximation. (13)The cumbersome "nonderegulated" term is used here in place of "regulated," because "regulated" and "state deregulated" are only two of several possible policy regimes. Fortunately, as late as 1984 locally regulated systems comprised 72.9 percent of all "non-deregulated" systems, while an additional 10 percent were subject to CPI or other local control, even while being listed under one of the three "local regulation" categories (DCA [1984, 24]). At any rate, the object of interest is not the absolute price level, but the first difference, so the "mixing" problem is not a severe one. The control group still picks up macro cable market trends in California to compare to first-year deregulated system changes. (14)I employ an Aspin Welch (t-statistic) test to evaluate a difference in means. This is defined as [Mathematical Expression Omitted] and [Mathematical Expression Omitted] (15)As the official listing does not separate the two types of deregulated systems, I identified systems with greater than 70 percent penetration as of the date of deregulation as partially deregulated. (16)There is additional evidence that deregulation leads to higher prices and higher penetration. While it is difficult to use cross-sectional models in predicting the price of cable, in response to a reviewer's suggestion I ran price and penetration equations using such explanatory variables as system size, system age, home density, availability of off-air television, number of basic channels, number of premium channels, and the price of premium channels, along with a dummy for regulated / deregulated, for the year 1985. The explanatory power of such models (looking at F, t, and [R.sup.2] values) is not powerful, but both the price and penetration equation dummy coefficients are, without exception, positive (and occasionally significant) across various specifications. This is entirely in line with the time series results presented herein. (17)The argument used by the cable industry in support of AB699 was perfectly consistent with the resulting evidence: "The cable industry is anxious to throw off the shackles of local rate regulation because it feels these limitations have inhibited growth" (Margulies [1979a, V12]). (18)For more than a two-year lag in rate-raising behavior, the question becomes: Why did the cable firm elect to "deregulate" in the first place? In that the action was costly, and that it was itself an act of defiance vis-a-vis local authority, it appears to stretch credibility to impute n-year lags due to such subtle gamesmanship. (19)Any price control regime may expect difficulty in controlling product quality; one added factor making this market exceptionally difficult to price control is the legal inability of local governments to exert virtually any authority over the cable operator's product (due largely to the First Amendment standing of cable operators since the late 1970s). The observed effects of price controls-lower nominal price with no accompanying increase in quantity demanded-suggest a situation where both price increase and product upgrade lags are induced by the control regime. As rate increases are more costly to institute with regulation, and political delays are now an input, firms partly adapt by delaying quality improvements. (This is particularly convenient where additional product services are continuously becoming available to suppliers, as occurred in the rapid extension of cable television channels throughout the sample period.) (20)Rent-sharing coalitions are increasingly seen as the moving force behind regulatory change (see Peltzman [1989]; Gilligan et al. [1989]; Hazlett [1990]). (21)The City of Los Angeles, which has (unsuccessfully) defended its prerogative to issue a monopoly franchise all the way to the U.S. Supreme Court (see Preferred Communications, Inc. v. City of Los Angeles, 754 F.2d 1396 [9th Cir. 1985], aff'd, 106 S. Ct. 1034[1986]), did not even seek to have potential franchisees bid prices at all: "The council will not be able to compare rates proposed by the bidders. In Los Angeles, a cable franchise is awarded and the Department of Transportation sets rates later" (Harris [1981]). (Of course, this post-contract price negotiation could itself have been evaded under AB699.) (22)This would not appear an outcome unique to this market. Peltzman [1989, 22] finds that "railroad regulation provides a good illustration of the spreading of rents to nonproducer groups. Producers got something -- protection from competition... Then these gains were partly shared with other groups through cross-subsidies." (23)Of course, a third dimension is introduced here, which would indicate a 3-D maximization problem for the rational regulator. For ease of presentation, the relevant trade-offs are considered in two successive planes. (24)Incumbent politicians are included here among the organized nonproducers.. (25)Actually, the industry's original proposal amounted to little more than naked deregulation. When Governor Jerry Brown, acting on advice from his counsel (who had been a "media access" lawyer), threatened to again veto the measure, a quid pro quo was written into the law by attorney Monroe Price, also a public access advocate. The consideration required was, essentially, the payment of the $.50/subscriber/year "deregulation fee" to go to public access/local origination subsidies. See Margulies [1979a; 1979b; 1979c]. (26) This enforcement mechanism, while relatively efficient for those who exercise it, cannot be employed frictionlessly. Due to the truncated property rights available to public sector agents, serious transactions costs will accompany such institutions, including lengthy delays. In other words, even where it would be in the interests of producers, organized nonproducers, and consumers to, say, allow a rate increase in exchange for additional channel offerings, political agents may not be able to consummate the Pareto efficient bargain in a timely fashion, owing to the ill-defined nature of public sector property rights. Such a circumstances would put a drag on nominal price increases without aiding economic efficiency. But, given the property rights regime, political decision makers could still perceive them as the low-cost solution.

REFERENCES

Becker, Gary. "A Theory of Competition Among Pressure Groups for Competition." Quarterly Journal of

Economics, August 1983, 371-400. Besen, Stanley M. and Robert W. Crandall. "The Deregulation of Cable Television." Law and

Contemporary Problems, Winter 1981, 77-124. Beutel, Philip A. "Uncovering Cities' Selection Criteria in Monopoly Franchising." Working paper,

Miami University, Ohio, 1989. California Public Broadcasting Commission [CPBC]. The Impact of AB699, Sacramento, Calif.: State of

California, April 1982a. ______. The Impact of AB699: Appendices. Sacramento, Calif.: State of California, April 1982b. Choi, Sung C. Introductory Applied Statistics in Science. New York: Prentice Hall, 1978. DeAlessi, Louis. "An Economic Analysis of Government Ownership and Regulation: Theory and Evidence

from the Electric Power Industry." Public Choice, Fall 1974, 1-42. Demsetz, Harold. "Why Regulate Utilities?" Journal of Law and Economics, April 1968, 55-65. Denn, Paul D. and C. L. Smit. The Consumer and Cable in California. Sacramento: California Public

Broadcasting Commission, 16 May 1983. Department of Consumer Affairs [DCA]. Rates for Basic Cable Television Service in California.

Sacramento, 1984. Fogarty, Joseph R. and Marcia Spielholz. "FCC Cable Jurisdiction: From Zero to Plenary in

Twenty-Five Years." Federal Communications Law Journal, January 1985, 113-29. Gilligan, Thomas W., William J. Marshall, and Barry R. Weingast. "Regulation and The Theory of

Legislative Choice: The Interstate Commerce Act of 1887." Journal of Law and Economics, April 19 89, 35-62. Goldberg, Victor. "Regulation and Administered Contracts." Bell Journal of Economics, Autumn 1976,

426-48. Harris, Kathryn. "City Trying to Pick a Cable Service Using Old, Incomplete Data." Los Angeles

Times, 30 August 1981, Sec. II-1, p. 3. Hazlett, Thomas W. "Competition v. Franchise Monopoly in Cable Television." Contemporary Policy

Issues, Spring 1986a, 80-97. ______. "Private Monopoly and the Public Interest: An Economic Analysis of the Cable Television

Franchise." University of Pennsylvania Law Review, July 1986b, 1335-409. ______. "Cabling America: Economic Forces in a Political World," in, Freedom in Broadcasting, edited

by C. Veljanovski. London: Institute for Economic Affairs, 1989, 208-23. ______. "The Rationality of U.S. Regulation of the Broadcast Spectrum." Journal of Law and

Economics, April 1990, 133-75. Jarrell, Gregg. "The Demand for State Regulation of the Electric Utility Industry." Journal of Law

and Economics, October 1978, 269-95. Kagan, Paul. The Kagan Census of Cable and Pay TV. Carmel, Calif.: Paul Kagan & Associates,

1978-1985. ______. Cable TV Financial Databook. Carmel, Calif.: Paul Kagan & Associates, 1985a. ______. Cable Television Franchising. Carmel, Calif.: Paul Kagan & Associates, 31 January 1987a. ______. Cable TV Financial Databook. Carmel, Calif.: Paul Kagan & Associates, June 1987b. ______. Cable Television Franchising. Carmel, Calif.: Paul Kagan & Associates, 30 November 1987c. Leffler, Keith B. "Ambiguous Changes in Product Quality." American Economic Review, December 1982,

956-67. Margulies, Lee. "Pending Bill Pulls Plug on Fee Controls." Los Angeles Times, 19 June 1979a, Sec.

VI, p. 12. ______. "Brown May Veto Cable Bill." Los Angeles Times, 23 August 1979b, Sec. IV, p. 32. ______. "New Cable Bill Widely Hailed." Los Angeles Times, 17 September 1979c, Sec. IV, p. 20. ______. "Cable TV: High Hopes for New Law." Los Angeles Times, 17 December 1979d, Sec. IV, pp. 31,

40. McChesney, Fred S. "Rent Creation and Rent Extraction in the Economic Theory of Regulation." Journal

of Legal Studies, January 1987, 101-18. Moore, Thomas G. "The Effectiveness of Regulation of Electric Utility Prices." Southern Economic

Journal, April 1970, 365-75. National Telecommunications and Information Administration [NTIA]. Video Program Distribution and

Cable Television: Current Policy Issues and Recommendations. Washington, D. C.: U. S. Dept. of

Commerce, NTIA Report 88-233, June 1988. Olmstead, Alan L. and Paul Rhode. "Rationing Without Government: The West Coast Gas Famine of 1920."

American Economic Review, December 1985, 1044-55. Peltzman, Sam. "Towards a More General Theory of Regulation." Journal of Law and Economics, August

1976, 211-40. ______. "The Economic Theory of Regulation After a Decade of Deregulation." Brookings Papers on

Economic Activity: Microeconomics, 1989, 1-41. Posner, Richard A. "Taxation by Regulation." Bell Journal of Economics and Management Science,

Spring 1971, 22-50. Soble, Ronald. "New Laws on Cable TV Considered." Los Angeles Times, 11 January 1982, Sec. II, p. 3. Stigler, George J. "The Theory of Economic Regulation." Bell Journal of Economics and Management

Science, Spring 1971, 3-21. Stigler, George J. and Claire Friedland. "What Can Regulators Regulate: The Case of Electricity."

Journal of Law and Economics, October 1962, 1-16. Webbink, Douglas. "The Impact of Regulation and Competition on Cable TV Service Prices, Channels and

Pay Tiers." Photocopy, Bureau of Economics, Federal Trade Commission, 2 May 1985. Zupan, Mark. "Franchise Bidding Competitions: How Well Do They Promote Efficiency in Cable TV

Markets?" Working Paper, University of Southern California, March 1987.

THOMAS W. HAZLETT, Associate Professor, University of California, Davis, Department of Agricultural Economics, and Visiting Scholar, Center for Telecommunications and Information Studies, Columbia University. The author wishes to thank Thomas E. Borcherding, Robert J. Michaels, Brooks Wilson, Mark Zupan and three anonymous referees for helpful analytical suggestions. John Mansell, Kathleen Schuler, and Michael Morris offered strategic assistance in data collection and interpretation. Myunghwan Kim and Hong Jin Kim are responsible for excellent research work. The Institute for Governmental Affairs provided partial funding for this study.