The Influence of Plan Demographics on Contribution Behavior of 401(k) Participants

• Date: 01 March 2017
• DOI: http://doi.org/10.1111/rmir.12069
• Published date: 01 March 2017
• Author: Youngkyun Park

Risk Management and Insurance Review

C

Risk Management and Insurance Review, 2017, Vol.20, No. 1, 7-35

DOI: 10.1111/rmir.12069

FEATURE ARTICLE

THE INFLUENCE OF PLAN DEMOGRAPHICS ON

CONTRIBUTION BEHAVIOR OF 401(K)PARTICIPANTS

Youngkyun Park

ABSTRACT

This article examines the influence of plan demographics, which is defined as

the salary and tenure composition of a plan, on contribution behavior of 401(k)

participants. Results using a large sample of 401(k) plans show that plan demo-

graphics influence participant contribution rates directly or through interaction

with individual demographic characteristics. A participant’s contribution rate

can be varied by 1.6–3.0 percentage points for each age between 25 and 64,

depending on what plan demographics she is exposed to. In addition, for the

decision to contribute the maximum allowed amount, plan demographics sig-

nificantly interact with individual demographic characteristics.

INTRODUCTION

While defined contribution (DC) plans (e.g., 401(k) plans) that load the burden of en-

suring adequate retirement savings on individual employees have become the primary

retirement vehicle, substantial interest has been placed in finding what determines par-

ticipant contribution rates. Most previous research has focused on examining the rela-

tionship between participant contribution rates and plan design features, which include

employer matching schemes, loan offerings, investment menus, and automatic features

(Choi, Laibson, and Madrian, 2004). However, less attention has been given to partici-

pants’ social contexts in a plan. In fact, social contexts that surround participants may

influence participants’ contribution rates, the influence that is called “contextual peer

effects” in the social interaction literature (e.g., Manski, 1993, 2000).1Usually, partici-

pants in a DC plan are not able to directly observe their coworkers’ contribution rates

unless coworkers share the information. However, even without direct observations

on coworkers’ contribution rates, participants’ contribution rates may be influenced by

coworkers in the same plan because the exogenous characteristics of coworkers can fa-

cilitate or constrain sharing information or knowledge about retirement savings among

participants.

Youngkyun Park is an Assistant Professor of Finance in the College of Business and Eco-

nomics, University of Idaho, Moscow, ID; phone: 208-885-7154; fax: 208-885-5347; email:

youngpark@uidaho.edu.

1Manski (1993, 2000) distinguishes contextual (or exogenous) peer effects from endogenous peer

effects. Contextual peer effects arise when an individual’s choice is influenced by exogenous

characteristics of her peers, such as socioeconomic status. In contrast, endogenous peer effects

arise when an individual’s choice is influenced by her peers’ choices or outcomes.

7

8RISK MANAGEMENT AND INSURANCE REVIEW

Among various exogenous characteristics of coworkers, this research focuses on plan

demographics that are defined as participants’ salary and tenure composition in a plan.

The plan demographics may relate to social influence on retirement savings through

information or knowledge sharing among participants. For example, those with a low

income or a short tenure are less likely to want to save for retirementthan others (Hilgert

et al., 2003; Huberman et al., 2007; Lusardi and Mitchell, 2007, 2011). If they make up the

majority of participants in a plan, the information or knowledge sharing on retirement

savings would be less active among the plan participants than otherwise. Plan demo-

graphics have been used in prior studies as an instrument or control variable to reflect

the potential influence of plan demographics on participants’ behavior. For example,

Duflo and Saez (2002) adopt plan demographics as an instrument to examine whether

or not to participate in a retirement plan. Huberman et al. (2007) use plan-level aggre-

gate variables (e.g., average compensation and tenure) as control variables to examine

the relationship between participants’ characteristics and contributions. However, little

research has been conducted to directly examine the relationship between plan demo-

graphics and participant contribution rates.

This research examines the influence of plan demographics (at a plan level) on partici-

pant contribution rates (at an individual level). It also investigates cross-level interaction

effects of plan demographics and individual demographic characteristics on participant

contribution rates (at an individual level). The cross-level interaction effects indicate the

extent to which the effects of a participant’s demographic characteristics on her contribu-

tion rates are varied with plan demographics that she is exposed to. Toanalyze the plan

demographic effects, we use a hierarchical linear model (HLM) because participants are

nested within a retirement plan and because an HLM is appropriate in analyzing data

that have a hierarchical structure (Bryk and Raudenbush, 1992; Hofmann, 1997).

From a large sample of 401(k) plans at year-end 2007, we find that plan demographics

have significant effects on participant contribution rates, while individual demographics

and plan design features are controlled for. For example, if a participant in a middle-

income and mid-tenure (MI-MT) plan works in a low-income and short-tenure (LI-ST)

plan, the participant would decrease her contribution rate by 0.98 percentage points; in

contrast, if the participant works in a high-income and long-tenure (HI-LT) plan, she

would increase her contribution rate by 0.80 percentage points.2We al so fi n d th a t pl an

demographics interact with individual demographic characteristics for participant con-

tribution rates. For example, for a $10,000 increase in salaries, a participant increases

her contribution rate by 0.43 percentage points when she is in an MI-MT or HI-LT plan.

However, when she is in an LI and ST plan, the same increase in salary raises her contri-

bution rate by 0.27 percentage points, 37 percent lower than the increase when she is in

an MI-MT or HI-LT plan. In addition, for the decision to max out employee contribution

($15,500 in 2007), we find significant interaction effects of plan demographics and indi-

vidual demographic characteristics. For a $10,000 increase in salaries, the probability of

2An LI-ST plan indicates one where 50 percent or more of participants have a salary of less than

or equal to $35,200 and tenure of shorter than or equal to 5.1 years; an MI-MT plan indicates

one where 50 percent or more of participants have a salary of $35,201−$63,600 and tenure of

5.2−12.4 years; an HI-LT plan indicates one where 50 percent or more of participants have a

salary of more than $63,600 and tenure of longer than 12.4 years. Plan classification will be

described in detail in the third section.