FAMILY FIRMS, BANK RELATIONSHIPS, AND FINANCIAL CONSTRAINTS: A COMPREHENSIVE SCORE CARD

• Date: 01 May 2019
• Published date: 01 May 2019
• Author: Robert M. Townsend,Jesús Saurina,Alexander Karaivanov
• DOI: http://doi.org/10.1111/iere.12362

INTERNATIONAL ECONOMIC REVIEW

Vol. 60, No. 2, May 2019 DOI: 10.1111/iere.12362

FAMILY FIRMS, BANK RELATIONSHIPS, AND FINANCIAL CONSTRAINTS:

A COMPREHENSIVE SCORE CARD∗

BYALEXANDER KARAIVANOV,JES ´

US SAURINA,AND ROBERT M. TOWNSEND1

Simon Fraser University, Canada; Bank of Spain,Spain; MIT, U.S.A.

We examine the effect of financial constraints on firm investment and cash flow. We combine data from

the Spanish Mercantile Registry and the Bank of Spain Credit Registry to classify firms according to whether

they are family-owned, not family-owned, or belong to a family-linked network of firms and according to their

number of banking relations (with none, one, or several banks). Our empirical strategy is structural, based on

a dynamic model solved numerically to generate the joint distribution of firm capital (size), investment, and

cash flow, both in cross sections and in panel data. We consider three alternative financial settings: saving only,

borrowing and lending, and moral hazard constrained state-contingent credit. We estimate each setting via

maximum likelihood and compare across these financial regimes. Based on the estimated financial regime, we

show that family firms, especially those belonging to networks based on ownership, are associated with a more

flexible market or contract environment and are less financially constrained than nonfamily firms. This result

survives stratifications of family and nonfamily firms by bank status, region, industry, and time period. Family

firms are better able to allocate funds and smooth investment across states of the world and over time, arguably

done informally or using the cash flow generated at the level of the network. We also validate our structural

approach by demonstrating that it performs well in traditional categories, by stratifying firms by size and age,

and find that smaller and younger firms are more constrained than larger and older firms.

1. INTRODUCTION

We focus on heterogeneity in financial constraints for firms, creating a comprehensive “score

card” from our findings. We integrate data and theory and analyze jointly data on firm invest-

ment, cash flow, and capital from two large comprehensive databases on Spanish nonfinancial

firms—the Spanish Mercantile Registry and the Bank of Spain Credit Registry (CIR). We then

structurally estimate and test the underlying financial regime, among a range of alternatives, so

as to be specific about (some of) the mechanisms determining the financial constraints, through

the lens of theory, from the least to most constrained: moral hazard (MH), borrowing/lending, to

saving only. We stratify the data into a large number of cells, by age and size, family ownership,

number of banking relations (with none, single, or several banks), region, industry, and three

different time periods. To our knowledge, such a comprehensive score-card approach, combin-

ing a variety of structural models with a number of key stratifications of the data, is unique. A

further advantage of our approach is that the models of financial constraints that we estimate

and test are dynamic; that is, we explicitly model and pay attention to the firms’ intertemporal

∗Manuscript received January 2017; revised June 2018.

1We are grateful to the editor, Jesus Fernandez-Villaverde, and two anonymous referees for their very helpful com-

ments. Research support from the National Science Foundation, the National Institute of Health, the John Templeton

Foundation, the Gates Foundation through the University of Chicago Consortium on Financial Systems and Poverty,

and the Social Sciences and Humanities Research Council of Canada is gratefully acknowledged. This article is the sole

responsibility of its authors, and the views presented here do not necessarily reflect those of Banco de Espana or the

Eurosystem. We thank V. Salas and K. Ueda for comments and suggestions on earlier versions. We also appreciate the

excellent comments from conference audiences at the Far East Meeting of the Econometric Society, the European Eco-

nomic Association, the Society for Economic Dynamics, and the Econometric Society World Congress. We are grateful

to S. Ruano for her valuable contributions to an earlier and significantly different version of the article and to J. Gal´

an

for his excellent research assistance with the current version. Please address correspondence to: Robert M. Townsend,

Department of Economics, MIT, 77 Mass, Ave., E52-538, Cambridge, MA 02139. E-mail: rtownsen@mit.edu.

547

C

(2018) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of Social

and Economic Research Association

548 KARAIVANOV,SAURINA,AND TOWNSEND

incentives to save and invest, for example, being constrained now versus constrained in the

future.

Our estimation results, using both cross-sectional and panel data, show that family firms data

are fit best by a less constrained financial regime than the corresponding data for nonfamily firms.

This points toward an advantage of family-based firm networks in accommodating investment

needs to cash flows obtained in different parts of the network. By shifting funds from firms

with weaker investment opportunities and more cash flow to other firms where the opposite

holds, family networks may surmount financial constraints that otherwise could put downward

pressure on firms’ investment levels and smoothing capabilities. We show that this result is robust

across firm age and size, time period, region, industry, and banking relationship strata. That

is, although there exist important differences in firm characteristics across these stratifications,

once we control for that by estimating within the respective subsamples, we still uncover the

same major pattern regarding family versus nonfamily firms.

Looking within the group of family firms, we find strong evidence that family-networked

firms (that is, firms in a family-based network constructed by ownership shares) face less strict

financing constraints compared to nonnetworked family firms (“pure family” firms, which are

individually or family owned but not owners of other firms). In terms of the estimated financial

regime, “pure family” firms tend to fall in between the nonfamily and family-networked firms

and are more similar to nonfamily firms than to family-networked firms.

We also find a clear pattern by firm age and size—estimating using data from larger and older

firms yields a less constrained best-fitting financial regime compared to using corresponding

data for small and young firms. These results are consistent with the previous findings from the

empirical literature that larger and older firms tend to feature less sensitive investment/cash flow

relationship compared to smaller and younger firms and thus validate our structural method.

Stratifying by firms’ banking status, our results depend on the firms’ size and age. In the

whole sample and among small and young firms, the estimated best-fitting financial regime

for unbanked firms is more constrained than the best-fitting regime for single-banked and

multibanked firms. In contrast, we find a different pattern within the category of large and old

firms—in the period 2001–2007, large and old unbanked firms are estimated to be less financially

constrained than single-banked or multibanked firms. A possible explanation is the different

composition by industry and/or that these firms may have saved or financed their way out of

the constraints. We report extensively on this heterogeneity and our attempts to control for

selection.

As noted at the outset, we use data from two large comprehensive databases on Spanish

nonfinancial firms. First, we use the Spanish Mercantile Registry to obtain balance sheets, profit

and loss accounts, information on financial/real flows (capital, investment, and cash flow), and

firm characteristics (ownership, age, etc.) Second, we use the Bank of Spain Credit Registry

(CIR) to classify firms according to whether they have borrowed from a single Spanish credit

institution, from several institutions, or from none.2Since the minimum reporting threshold

was low (6,000 euros per loan) over our sample period, the CIR data are essentially a census of

all banking loans granted to nonfinancial firms in Spain. In the estimation, as in the model, we

focus on firms that continuously maintain the same banking status (unbanked, single-banked,

or multibanked) in all years of data.3

We construct an indicator of whether a firm is family-owned, not family-owned, or part

of a family network defined by ownership shares obtained from the firm registry. We first

identify all firms that are directly family-owned (50% or larger share is held by an individual

or family). We then add to this set of family firms all other firms owned with 50% or larger

share by the family firms in the initial set. We repeat the process until no new firms are added

2Jim´

enez et al. (2012, 2014, 2017) have used extensively both databases to analyze monetary policy and financial

stability. Here, we focus on the real side of the economy by investigating nonfinancial firms’ investment.

3We implicitly assume that multibank lending delivers an outcome as if a firm were dealing with a unified sector, as

when banks can coordinate, but this is a limitation of the model.

FAMILY FIRMS’FINANCIAL CONSTRAINTS 549

at further iterations. The end result is all firms controlled by an individual or family, directly

or indirectly. We further subdivide family firms into two types: “pure family” firms (directly

owned by individuals or families and which do not own other firms) and “family-networked”

firms (part of a network of firms constructed by ownership). The rationale is to distinguish

different types of family firms, for example, a “mom-and-pop store” versus a company within a

large business group.

In more detail, we analyze the investment behavior of Spanish firms taking into account

their age, size, banking relationships, and ownership structure. We also stratify the firms by

region (Madrid, the Mediterranean, and the Ebro river valley vs. the rest of the country) and

by industry (construction and real estate vs. all other nonfinancial sectors). We analyze firm

investment in a country where business funding is overwhelmingly dominated by banks and in

a period that corresponds to the longest continuous expansion of the Spanish economy under

a lending boom that fueled firms’ investment expansion. We analyze three subperiods in a

balanced manner: 1997–2000, which corresponds to the investment expansion after the 1993

recession in the wake of entering the eurozone; 2001–2003, an economic slowdown as a result of

recession in core eurozone countries; and finally 2004–2007, the years right before the financial

crisis and also a time when lending and liquidity were at their peak. Although the years that we

study, 1997–2007, encompass different growth periods of the Spanish economy, we do not find

a strong pattern in terms of the best-fitting models of financial constraints over time, although

we do find more instances of the least constrained (MH) regime estimated as best fitting in the

2004–07 data, the period of the credit ramp-up before the financial crisis.

Our definition of financial constraints is about how damaging, or alternatively how flexible, is

the financing/information regime in which a firm resides, based on the estimation of alternative

dynamic models. The financial regimes range from saving only to unrestricted borrowing/lending

with an institution or market to MH constrained insurance and credit arrangement. We tabulate

the placement of firms into different cells based on key characteristics emphasized in the litera-

ture: size and age, bank finance (no, single, or multiple bank lenders), family status (nonfamily,

family-owned, in a family-linked network determined by ownership), region, industry, and time

period. We further disaggregate the data by intersecting these categories, for example, small

and young single-banked firms that are family-connected versus small and young single-banked

with no family connections, etc. The bottom line is a score card or an indicator for which group

of firms is more severely financially constrained and which less so.

A large literature on firm investment at the micro level studies the role of financial con-

straints.4Some authors use models of information or incentive problems in capital markets

to motivate the role and origin of endogenous financial constraints.5Other authors assume

exogenously incomplete credit markets, for example, borrowing and saving in a single asset.

In both cases, the ultimate result is limited access to external finance—firms have to restrict

investment when internal cash is insufficient to invest at the first-best level.6Compared to pre-

vious approaches relying on more reduced-form analysis, including using panel data,7the main

advantage of our structural approach is that we can not only assess whether financing constraints

affecting firm investment are absent or present (perfect vs. imperfect credit markets), but also

determine the most likely nature of the financial constraints within the set of structural mod-

els we compute, estimate, and statistically test. Here, we feature three prototypical models of

exogenously or endogenously incomplete markets, but, as shown in Karaivanov and Townsend

4Schiantarelli (1996), Hubbard (1998), and Strebulaev and Whited (2012) provide extensive reviews.

5On adverse selection, see Jaffee and Russell (1976) or Stiglitz and Weis (1981). Myers and Majluf (1984) focus on

information problems affecting equity financing. The effects of MH are treated in Jensen and Meckling (1976) among

others. Williamson (1987) derives the possibility of credit rationing in the context of optimally designed contracts when

profit outcomes can only be observed at a cost. On costly state verification, see also Townsend (1979) and Diamond

(1984).

6The cost gap between external and internal funds could be explained by information frictions but could also be due

to taxes or other transaction costs.

7See, among others, Bond and Meghir (1994) and Bond et al. (2003).