Consumer Behavior in Mexico’s Privatized Social Security System Justine Hastings, Yale University and NBER Motivation Movement towards increased choice, privatization in traditionally publicly provided markets Goals of harnessing market power to improve efficiency (education, healthcare, social security) Efficiency gains depend on consumer behavior Full-information Frictionless market Forward-looking, rational decision making Look to private markets, we find that many of these assumptions fail to hold, competitive markets don’t always provide efficient solution Competition on advertising may replace price competition if consumers focus on brand-name more than price (Dorfman and Steiner (1954), Dixit and Norman (1978), Sirri and Tufano (1998), Barber, Odean and Zheng (2005), Cronqvist (2006) and Bagwell (2007)). Consumers may focus more on brand name or other salient features if prices or their relevance are difficult to obtain or understand (Hastings and Tejeda-Ashton (2008), Hastings and Weinstein (2008), Kling et al. (2008), Abaluck and Gruber (2009), DellaVigna (2009)) Competitive firms may not have incentive to educate them (Ellison (2006), Gabaix and Laibson (2006)) May rely on peers for guidance (Duflo and Saez (2003)), leading to network externalities which raise barriers to entry (Katz and Shapiro (1994)) Switching costs can lead to suboptimal behavior, dampen price competiton (Klemperer (1995), Farrell and Klemperer (2007); Madrian and Shea (2001), Beshears et al. (2006), and Carroll et al. (2009)) Goals of Project Use detailed administrative data from Mexican Social Security System to examine How consumers decide to manage their private pension accounts Test how switching costs, peers, advertising, government information effect account management decisions Draw implications for price competition, government design of markets, directions for future research Mexico’s Privatized Social Security Market Privatized market in 1997, established private accounts Market during our time period – 2004-2006 Approved fund providers (Afores) ; 13 – 21 in operation in the market Regulated funds (2 Siefores each; no persistent market outperformance) Centralized administration to decrease fixed costs Contributions of 6.5% of wage, accrues in account until retirement Affiliates can choose an Afore, switch with low cost between them Afores charged 2 fees: flow fee and balance fee Early withdrawal for unemployment insurance, expense of marriage No fee for moving accounts between Afores Regulator published the CEF (CONSAR equivalent fee) that distilled these 2 fees into one fee statistic Afores hire sales agents to visit/call workers to convince them to switch their account Data and Empirical Approach Data on all accounts Employment and contributions – wages, employer id, contributions to SAR account, Afore Switches – switches between Afores, liquidation amounts and dates Gender, date of birth, and zip code of residence (in June 2006) Empirical Approach Analyze determinants of when workers switch accounts between Afores Discrete time hazard model; logistic regression Random utility model of Afore choice Estimate importance of switching costs, peers, brand name and advertising, management fees Identification Switching costs – examine impact of exiting, entering labor force and switching jobs Peers – construct share of each Afore at place of work. Fees (flow and balance) Regulatory changes in tenure discounts Variation in costs across people and within people as employment status changes Fee changes conditional on Afore fixed-effects; entry of new Afores CEF What is the effect of Afore share on management behavior? Examine effects of switching jobs to place where your Afore is in the minority Changes in assumptions that are used to calculate CEF Advertising Sales force exposure; control function approach to identify effect off of factors that shift costs of sales force but not demand for Afores (Rivers & Vuong (1998), Villas-Boas (1999), Petrin and Train (2006)) Specific Findings Timing of account switching: Overall, people do little to manage their accounts, switching only once every 5-6 years on average. Switching costs are very important: Salience is very important: Exiting the formal sector decreases probability of switching by 82-86% Peers/employers are very important for lower-income workers Entering period for filing for Unemployment Insurance increases probability of switching by 370% Changing job to an employer where current Afore is less popular has same impact as direct marketing exposure Changes in prices are relatively ineffective in moving demand forward in time Implies price cuts are not profitable except for smallest firms Specific Findings Timing of account switching: (discrete-time hazard model) Overall, people do little to manage their accounts, switching only once every 5-6 years on average. Workers switch very infrequently; increasing in income Switching costs are very important: Exiting the formal sector decreases probability of switching by 82-86% Peers/employers are very important for lower-income workers Entering period for filing for Unemployment Insurance increases probability of switching by 370% Salience is very important: Lowest wage earners, 0.002; Highest wage earners, 0.0215 Changing job to an employer where current Afore is less popular has same impact as direct marketing exposure Changes in prices are relatively ineffective in moving demand forward in time Implies price cuts are not profitable except for smallest firms Specific Findings What determines Afore choice: (random utility model) Pessimistic findings: demand is price inelastic Less elastic to harder-to-determine actual cost More elastic to government sponsored measure of fees Both are increasing in income; but not by much Optimistic findings Low income workers focus on peers, brand name High income workers more likely to chase past returns (even though Afores do not display persistent market outperformance) Investors are responsive to advertising; particularly among low income workers Requirements that switching accounts sign that they have viewed CEF table has positive impact on elasticity w.r.t. CEF – increases sensitivity by 20% Sales agents decrease dependence on peers However, advertising also increases responsiveness to 12 months past returns Suggests that Regulation on information; marketing can improve price sensitivity at time of choice Coordinating market, designing accounts with periodic immediate usefulness of savings can increase demand elasticity over time. Concluding thoughts; future research Consumer behavior mitigates price competition in private accounts market Inattention to accounts with subscription good market Consumers are not price sensitive; behave as if information costs are high generating network effects and competition on advertising Absent threat of regulation, prices would have been higher Suggests that market design and marketing can have big benefits Coordinating demand side, allowing for early contact with account can increase demand elasticity significantly Lowering switching costs may also have an impact Information framing and government provision of information may really move demand Hastings & Tejeda-Ashton (2008) Field experiments in Mexico and Chile Moving part of the market towards price elasticity may have large equilibrium effects