Essays on firm growth and financing
- Autores: Fernando Riveiro Formoso
- Directores de la Tesis: Andrés Erosa Etchebehere (dir. tes.)
- Lectura: En la Universidad Carlos III de Madrid ( España ) en 2023
- Idioma: inglés
- Tribunal Calificador de la Tesis: Árpád Ábrahám (presid.), Felix Maximilian Wellschmied (secret.), Alexander Monge Naranjo (voc.)
- Programa de doctorado: Programa de Doctorado en Economía por la Universidad Carlos III de Madrid
- Materias:
- Enlaces
- Tesis en acceso abierto en: e-Archivo
- Resumen
The unifying theme of this dissertation is the behaviour of firms over their life cycle, namely regarding their access to financial resources and their growth in terms of size, and the implications firm growth and financing have from an aggregate perspective. An issue that has been of interest to the macroeconomics literature is: how should financial resources be allocated across firms? (Hsieh and Klenow, 2009; Restuccia and Rogerson, 2008). This dissertation analyses how capital should be allocated to young firms over their first years of activity. On one side, young firms are often small, which may prevent them from accessing financial markets (Gopinath et al., 2017; Moll, 2014; Ruiz-García, 2021). On the other side, in some sectors, there is high uncertainty regarding the underlying quality of young firms (Kaplan and Strömberg, 2004). These issues might deter firms that have a high productivity (or are potentially highly productive) from growing over life. Those highly productive firms are worth financing from an aggregate perspective. As a consequence, exploring the conditions under which productive young firms can flourish is a first-order issue to promote a well-functioning productive sector and a good economy-wide performance. Chapters 1 and 2 of this thesis explore the nature of firm growth and the life-cycle dynamics of firms that can potentially grow strongly over life, but are subject to borrowing constraints. Chapter 3 analyses the role that learning has on investment and exit decisions of high-risk, young companies.
In the first chapter, "A Life-Cycle Study of Productivity and Factor Allocation of Spanish Firms", the nature of firm growth in Spain and the existence of distortions that may prevent young Spanish firms from growing are empirically studied. This chapter is motivated by several macroeconomic and firm-level facts documented for the Spanish economy. On one side, the capital misallocation literature for Spain has documented a poor aggregate performance of the Spanish economy over the recent decades, which is considered to be related to a bad allocation of resources across heterogeneous firms, and to the presence of financial constraints affecting small firms (Fu and Moral-Benito, 2018; García-Posada and Mora-Sanguinetti, 2014; García-Santana et al., 2020; Gopinath et al., 2017). On the other side, Spanish firms are typically small and display low growth over their life cycle (Hsieh and Klenow, 2014; Rubini et al., 2012).
I look at these motivating facts by adopting a firm-life-cycle perspective. Indeed, in a recent study for the United States, Pugsley et al. (2021) find that a small fraction of firms are displaying very high growth over their first years of activity, and that this small fraction of firms ends up becoming very important for the aggregate US economy. I study whether this high-growth phenomenon could be taking place in Spain; and if not, what are the reasons why we do not observe those high-growing firms.
I use micro data on Spanish companies to study the life-cycle behaviour of firm productivity and factor inputs. The data at hand is particularly well-suited for performing an empirical study on firm growth and factors hindering it, for it has two distinctive features. First, the dataset I use contains information on small firms -- i.e. it is not limited to listed firms. Thus, I can track the life cycle of firms that are characteristic of the Spanish economy. Second, the dataset is not limited to administrative employment data and consists of the balance sheets of Spanish companies. Therefore, I can study not only a measure of employment at the firm level, but I can also examine firm-level value added and the book value of long-term assets, i.e. measures for output and capital. This information allows me to pin down a measure of firm-level total factor productivity (TFP).
With this information, I perform a life-cycle study of the evolution of firm-level TFP, employment and capital. The study consists of two steps. First, I use the methodology proposed by Pugsley et al. (2021) to study whether the TFP data supports the idea that some firms are more determined to grow than others at birth, in terms of their productivity. Second, I use the data on TFP and factor inputs to study how well resources are allocated to Spanish firms over their life cycle. I study whether the manner capital and employment change over the firm life cycle can inform about factors generating resource misallocation.
Regarding the first part of the study, I find evidence that some firms are expected to grow strongly when they are young, in terms of productivity. I use the life-cycle TFP data (namely, the empirical autocorrelation by age of the logarithm of firm-level total factor productivity) to estimate a firm-level productivity process. This process considers that productivity over the life cycle has two determinants: first, ex-post shocks that realise over life, as it is usually considered in the firm dynamics literature; and second, an ex-ante component embedding the idea that some firms are more determined to grow than others at birth. The evolution of the ex-ante component, in turn, is characterised by a firm-specific initial condition and a firm-specific long-run steady state, which give room to differences in expected productivity growth rates across firms. The empirical evidence supports the idea that ex-ante components in the productivity process are relevant for explaining the life-cycle dynamics of firm-level total factor productivity in the data, and that some firms in Spain have a high growth potential at birth.
With respect to the second part of the study, I find that the data on factor inputs points to the existence of firm-level frictions affecting resource allocation to firms over life. By studying the sample of Spanish companies, I observe that empirical autocorrelations by age (which describe the manner variables correlate with themselves for different age pairs over the firm life cycle) differ across three variables of interest -- namely, logged TFP, employment and capital. I argue that these empirical differences cannot be rationalised by a completely frictionless model of the firm, where the evolution of factors of production over the firm life cycle perfectly maps that of firm-level productivity. Indeed, in the data, logged capital and employment are more autocorrelated that logged TFP over the firm life cycle. This indicates that real-life firms do not live in a frictionless world, and suggests that factors of production are more difficult to adjust over the life cycle than firm-level productivity. Therefore, empirical autocorrelations by age of factor inputs are shown to be useful to identify firm-level frictions. In particular, they can be used to pin down information about borrowing constraints, which the capital misallocation literature has pointed as the main culprit of the poor resource allocation and productivity performance in Spain.
In the second chapter, "Factor Misallocation and High-Growth Firms in Spain", I build a firm-dynamics, general-equilibrium model to study whether firms with a high growth potential at birth manage to realise high growth effectively, or whether they are deterred from growing by the existence of firm-level frictions, namely borrowing constraints. This chapter is motivated by the empirical findings in Chapter 1 regarding high-growth firms and firm-level frictions, and addresses the firm-level effects of borrowing constraints and their aggregate impact on the Spanish economy.
I model a stationary economy in which firms are born with heterogeneous initial capital endowments, produce over the life cycle, and exit. The model has two key elements: a range of firm-level frictions that are important at the firm level (Cooper and Ejarque, 2003; Cooper and Haltiwanger, 2006; García-Posada and Mora-Sanguinetti, 2014), and a firm-level productivity process in the spirit of Pugsley et al. (2021). First, firms choose how much employment to hire and how much to invest subject to adjustment costs in capital, and a borrowing constraint that depends on the amount of capital owned. The borrowing constraint restricts firm investment, and also labour expenditures in a working-capital-constraint fashion. Second, I feed the productivity process estimated in Chapter 1 into the model. This process embeds the idea that some firms have a high growth potential, and it is motivated by my empirical findings on the life-cycle behaviour of firm-level TFP. Importantly, this process has not been previously considered in the literature of financial frictions.
I calibrate the model to match the life-cycle empirical evidence on autocorrelations of logged employment and capital by age, which has been discussed in Chapter 1. The model provides a good fit of the Spanish data. In addition, the model is capable of capturing some relevant untargeted dimensions -- namely, the average employment by age of Spanish firms over the first ten years of activity.
By simulating the calibrated general-equilibrium model, I arrive at the following micro-level and macro-level results. First, regarding firm-level results, I find that 1.1% of simulated firms are classified as having a high growth potential (¿high-growth-potential firms¿). I find that 6.5% of total firms realise sufficiently high growth in terms of simulated employment -- those are labeled ``effective high-growth firms''. I find that simulated firms that effectively realise high employment growth own on average a notably high initial capital endowment. This suggests that firms with lower borrowing constraints are able to grow more, for they overcome the friction early in life. Indeed, I find that only 36.6% of high-growth-potential firms effectively realise a sufficiently high employment growth, which suggests that they are damaged by borrowing constraints. With respect to macro-level results, I find that, although high-growth-potential firms account for only 1.1% of simulated firms, they account for 3.3% of aggregate TFP and 7.1% of aggregate output in the simulated economy.
Next, I use my model to perform the main counterfactual exercise of the chapter, which consists of removing borrowing constraints affecting Spanish firms. When I do so, the aggregate effects are striking. Aggregate output increases by 22.9% and aggregate TFP increases by 20.5%. This indicates that, when financial frictions disappear, there is a substantial resource reallocation towards most productive firms. In particular, the fraction of high-growth-potential firms that effectively realise high growth increases from 36.6% to 76.2%, and their aggregate relevance is also larger when constraints are eliminated. From this exercise, we learn that high-growth-potential firms are deterred from growing due to borrowing constraints and that they can make a sizable aggregate impact to the economy when they are not constrained.
I argue that explicitly considering life-cycle components in the firm-level productivity process (i.e. accounting for the phenomenon of high-growth-potential firms) is important to better understand the aggregate effects of borrowing constraints. To see this, I consider what would happen if we abstract from the phenomenon of high-growth-potential firms. I solve and simulate an alternative model that considers a standard productivity process (i.e., without heterogeneity in expected growth rates) similar to those often used in the financial frictions literature (Buera and Shin, 2011; Moll, 2014). This alternative firm-dynamics model with the standard productivity process fed in is calibrated to match the same moments as before. When I simulate the alternative model, my quantitative results are different from those from the benchmark model with heterogeneity in expected productivity growth rates. Under the alternative, more standard model, there are no high-growth-potential firms and simulated firms are on average growing less than in the benchmark model. In spite of this, removing borrowing constraints generates smaller aggregate effects: 14.07% increase in aggregate TFP (vs. 20.5% in the benchmark model) and 13.59% increase in aggregate output (vs. 22.9%).
This exercise sheds light on the mechanism why firm heterogeneity in expected growth rates is important for understanding the aggregate effects of financial frictions. If a firm has a high growth potential, it will have a low productivity at birth, but it will become highly productive after some periods of life. During the first years of life, this firm will earn low profits and will thus perform low investments. After some years, this firm will have attained its potential productivity, but it will be borrowing constrained and thus inefficiently small. Therefore, high-growth-potential firms benefit a lot from the removal of financial frictions after some periods of life, and this generates sizable aggregate effects. This mechanism is not present if we consider a more standard productivity process, as in Moll (2014) and Buera and Shin (2011), and thus steady-state aggregate effects from borrowing constraints are larger in a setting that accounts for life-cycle components in the firm-level productivity process.
In the third chapter, "Venture Capital Investments and Learning over the Life Cycle", I study how young, high-risky firms make investment, liquidation and sale decisions. The issue of how to allocate financial resources to young firms is crucial when we talk about innovative, high-uncertainty entrepreneurial projects, whose underlying quality and future prospects are unknown at birth. Typically, when thinking about the financing of innovation, the financial contracting literature studies the agency problem between an entrepreneur and a financier, thus being concerned with the optimal contract design (Bergemann and Hege, 1998; Marx, 1998; Repullo and Suarez, 2004; Schmidt, 2003). Differently from these papers, the objective of this chapter is to shed light on a feature of the firm life cycle that has not been studied in much detail in the context of high-risk firms and their financing: their ability to learn about the company's uncertain quality, as firm-level results are observed over time. This chapter studies how learning affects life-cycle investment and exit decisions, as well as firm value, in contexts of high uncertainty.
I develop a model of the firm that imitates realistic features of young, high-risk companies, such as uncertainty about a firm's own quality, staged financing, exit strategies, and the realisation of period cash-flows that yield information about the firm's unobserved quality. In the model, an agent is born owning a firm whose quality is unknown, and thus forms beliefs about it. The original owner of the firm is the only agent in the economy that has the ability to exert growth investments to improve her project. Upon having liquidated her firm or having sold it to the market, no more investments can be made into the firm. A key feature of the model is that, at every period of firm life, the agent receives period cash-flows that convey information about the true quality of the project. As far as these signals are sufficiently informative about the underlying quality, the firm owner can learn and update her beliefs, which in turn may change her discrete choice to keep her firm, to sell it to the market, or to terminate it.
By numerically solving the model and simulating it, I show that this framework where firms learn about their uncertain quality captures documented patterns in the venture capital literature, such as delayed exits and contingent stage financing (Kaplan and Strömberg, 2003; Metrick and Yasuda, 2010). First, I find that the ability to learn, jointly with the capability of terminating the project at every period, gives the agent an option value of waiting and updating her beliefs. This is possible if cash-flows are informative, to some extent, about the true quality of the project. The option value of not terminating, jointly with the cost of selling the project (e.g. IPO costs from taking the firm to the market), gives incentives for owners of young, high-uncertainty companies to keep their firms and exert growth investments over the life cycle. These incentives do not exist when firms are sufficiently old and there is no uncertainty regarding their true quality.
Second, I find that the ability to learn from period informative signals renders optimal investment decisions contingent on period cash-flow realisations, which is consistent with documented patterns at Kaplan and Strömberg (2003). In the simulated data, I find that period investments and cash-flow present a high contemporaneous correlation, of 0.75, which is induced by the learning mechanism. Should we turn signals into completely uninformative ones (so that there is no learning), optimal investment would be unaffected by realised cash-flows.
Finally, I show that the noise of period cash-flows matters for determining firm value. In particular, when signals are extremely noisy, so that no information can be inferred about the true quality of firms, no owner would be willing to keep her project. Nevertheless, if the ability to learn from cash-flows is sufficiently high, then keeping and investing provides value to the owner of an uncertain firm. I find that firm value increases in the informativeness of the signal, and this is particularly so for firms whose initial uncertainty at birth is higher. The model thus provides reasons why a superior ability to learn about firm-level results over the life cycle is a value-adding skill that is particularly interesting for innovative, high-risk entrepreneurial projects.
