Bailing on the Car That Was Not Bailed Out: Bounding Consumer Reactions to Financial Distress

• Date: 01 June 2017
• Author: Cristian Huse,Nikita Koptyug
• DOI: http://doi.org/10.1111/jems.12184
• Published date: 01 June 2017

Bailing on the Car That Was Not Bailed Out:

Bounding Consumer Reactions to Financial Distress

CRISTIAN HUSE

Stockholm School of Economics

Box 6501 113 83 Stockholm Sweden

cristian.huse@gmail.com

NIKITA KOPTYUG

Research Institute of Industrial Economics

Grevgatan 34 Box 55665 102 15 Stockholm Sweden

nikita.koptyug@ifn.se

We examine how consumers react to the financial distress of durable goods manufacturers by

studying the Swedish new car market. We employ a difference-in-differences matching method-

ology whereby we compare sales of carmaker Saab with those of a control group of substitute

products. To account for possible substitution between products in the treatment and control

groups, we propose and apply bounds to our difference-in-differences matching estimator. We

then refine the bounds and provide conditions under which they depend only on product elas-

ticities. We find that there was a significant decrease in the sales of Saab following its filing for

administration.

1. Introduction

In a number of industries such as the automobile industry,both warranties and post-sales

assistance are prevalent and of such great importance to consumers that these are yet

another dimension over which carmakers differentiate their products (Cars.com, 2011a,

b). As a result, whenever a company producing automobiles, or any other durable

good, faces financial difficulties—bankruptcy in the extreme case—one might expect

consumers to shy away from its products due to risks associated with the company

ceasing to exist and post-sales services tied to the product becoming worthless (as

modeled in, e.g., Titman, 1984).

The importance of post-sales services to consumers make the auto industry well

suited for studying how consumers perceive and react to the financial difficulties of

carmakers. The studies by Hammond (2013a) and Hortac¸su et al. (2013) have made

use of these features and, perhaps unsurprisingly, found that that financial distress of

a carmaker does affect the secondary car market; on average, consumers are no longer

willing to pay as much for a used car produced by a carmaker that starts experiencing

financial difficulties.

Nikita Koptyug gratefully acknowledges financial support from the Foundation of Banking Research, the

Jan Wallander and Tom Hedelius Foundation and the Marianne and Marcus Wallenberg Foundation. We

thank Tobias Olsson at Vroom Stockholm for making data available and participants at the EARIE, Jornadas

de Economia Industrial, IIOC, MOITA Workshop, SBE, and SSE (Empirical Micro Workshop,Economics and

FinanceLunch Seminars) for comments. We are also indebted to Laurent Bach, ´

AureoDe Paula, Bob Hammond,

Erik Lindqvist, Anindya Sen as well as our discussants C´

eline Bonnet, Rodrigo Moita, Leonard Nakamura,

and Frank Verboven for comments. Lastly, we are grateful to the editor, coeditor, and an anonymous referee

for useful comments and suggestions.

C2016Wiley Periodicals, Inc.

Journal of Economics & Management Strategy, Volume26, Number 2, Summer 2017, 337–374

338 Journal of Economics & Management Strategy

To study how consumers react to financial difficulties and distress on the primary

market, one has to establish among other things whether it is a carmaker’s financial

health that is affecting consumer behavior or vice versa. In this study,we aim to resolve

the issue of reverse causality by focusing on the effects of Swedish carmaker Saab’s

financial distress and filing for administration on the Swedish new passenger car mar-

ket.1Saab, originally a domestic Swedish carmaker whose products were praised by

their admirers for the fine engineering and characteristic design, was facing financial

ills due to the parent company’s General Motors’ (GM) own financial problems of late

2008.2Following the financial aid package provided by the U.S. Government to bail

out GM in late 2008, it eventually became clear that Saab would not benefit from a res-

cue of its parent company. The terms of the emergency loan granted to GM were such

that the money was to be used to save American jobs and other funds would have to

be used to keep Saab operational (SvD, 2008b). Due to these circumstances, we argue

that Saab’s financial situation was not a concern for potential buyers until GM started

experiencing its own financial troubles. Having lost the backing of GM in early 2009

after a strategic review, the financial difficulties facing Saab led the company to file for

administration in February 2009. It was only months later that GM agreed on Saab’s sale

to a new owner with the financial backing of both the Swedish Government and the

European Investment Bank (EIB). As a result, once a binding deal regarding the sale of

Saab to Koenigsegg Group was signed in August 2009, Saab’s administration came to

an end.

Empirical Strategy. We use vehicle registration-based data to study purchases of

new Saab cars by private individuals in Sweden during the years 2006–2009. By doing

so, we contribute to a better overall understanding of how consumers react to financial

distress and on the impact of financial distress on a company.3

In our analysis, we take advantage of a key institutional setting of the Swedish

new car market: Sweden being a small market, car dealers keep low inventory levels,

so consumers typically order a car a few months in advance. This means that as long as

Saab was producing cars, it would have been possible to order a car as before. Further,

this results in very few episodes of sales or rebates from the part of carmakers or dealers,

making the list prices, we observe a much improved approximation to transaction prices

than in most markets, for example, the United States. As a result, we are able to make

use of both price and quantity information in our analysis.

Our identification relies on the credible exogeneity of GM’s financial health to

Saab product market performance but not to its financial one. That is, prior to filing for

administration, Saab could count on the financial backing of GM, a large multinational

conglomerate that did on several occasions inject funds into its subsidiary. However,

this situation changed abruptly once GM applied for financial assistance from the U.S.

government, which set as a precondition that U.S. funds are only used to keep U.S. jobs

and could not be used to support Saab or other foreign brands.

To identify and quantify the impact of Saab’s financial distress on consumer pur-

chasing behavior, we use a difference-in-differences (DD) matching estimator whereby

we compare the sales of Saab, the treatment group, to those of comparable cars on

1. Saab is known among car enthusiasts for (among other things) launching the first mainstream tur-

bocharged car ever sold, the Saab 99 (Popular Mechanics, 2008).

2. As detailed below, supply-side-relatedconcerns were not a major concern in Saab’s case.

3. Although companies and the Swedish government purchase a significant share of new cars in Sweden,

they do not necessarily have the same objectives as consumers when purchasing automobiles. For instance,

the Swedish government has a mandate to buy environmentally friendly vehicles for non-emergency services.

Consumer Reactions to Financial Distress 339

the Swedish market, the control group. We propose constructing our control group by

matching Saab products to a set of close competitors according to observable prod-

uct characteristics.4More importantly, we propose a way to estimate treatment effects

when both treated and untreated subjects are likely to be affected by the treatment.

To do so, we propose economically-motivated bounds for the DD estimator we con-

sider. The motivation for the bounds comes from the fact that our control group for

Saab products consists of close substitutes to the treated subjects. The fact that there

may be substitution from Saab products to the control group renders standard methods

used in the treatment effects literature unfit for our purposes (see Angrist and Pischke,

2010, and Nevo and Whinston, 2010, for a closely related exchange). For instance, in

our application, substitution from the treatment to control group results in standard

DD estimates being biased upward (in absolute terms), that is, any consumer reaction

to Saab’s financial ills is likely to be overstated by the standard DD estimator. To ac-

count for this, we propose bounds for our DD matching estimator by assuming that

the true effect lies between no- and perfect substitution between treatment and control

groups.

One concern raised by our empirical strategy—the matching procedure in

particular—regards how the size and composition of the control group affectthe bounds

we propose. That is, the method might give researchers excessive leeway, so that in the

extreme case, the choice of the control group may determine the bounds to the DD effect.

To address such concern, we provide conditions under which our bounds are either not

substantially, or not at all, influenced by the choice of control group. First, a homogene-

ity condition whereby cross-price elasticities between the treated subject and products

within a market segment are similar. Second, a dominance condition according to which

the own-price elasticity of the treated subject dominates cross-price elasticities between

itself and products in the same market segment, which, in turn, dominate cross-price

elasticities between the treated subject and products in other market segments. Third, a

symmetry condition whereby products in the treatment and control groups have sales of

similar magnitudes. Under these conditions, the bounds become a function of own- and

within-group cross-price elasticities of the marketed products. Such conditions are con-

sistent with empirical findings for differentiated product markets (see online Appendix

B for an illustration) and allow the method to be applied within Industrial Organization

and Antitrust settings.

Although one could think that the assumptions we impose would result in un-

informative (i.e., overly wide) bounds, we show in our application that even without

refining the bounds, we obtain results pointing to a significant effect of Saab’s financial

distress on its sales.

Main Findings. In contrast to previous findings for the used car market, we find

limited evidence that prices of the Saab brand as a whole responded to its financial

distress. As a result, we focus primarily on its sales. Doing so, we find ample evidence

that consumers react to Saab’s financial troubles; our most conservative results point

to consumer reactions corresponding to a decrease of over 200 units sold by the Saab

brand, as compared to the control group, a sizable 50% and 56% effect if compared

to unconditional average sales of Saab which amount to 470 units per month. Given

the ample evidence of the existence of a home-bias in both the International Trade and

4. We also provide robustnesschecks using the synthetic matching methods of Abadie and Gardeazabal

(2003).