Periodic Reporting for period 3 - FIRMNET (Firms and Their Networks)
Berichtszeitraum: 2020-09-01 bis 2022-02-28
Two very recent events – the Trump-led anti-globalization policies (with the associated tariffs war or skirmishes against China or Europe) and the Covid19 pandemics – have placed the global fragmentation of the production value chains center stage. Firms and their Networks, the ERC project about which I report here, by taking stock of this fragmentation is trying to understand how firm-to-firm as well as firm-to-workers’ relations hold on a day by day basis as well as in the face of uncertainty. Economic shocks can have an economic as well as a technological origin, such as the creation of products or of services that become available to firms. The various types of networks between workers and firms, based on social connections or mere employment relations, as well as between two (or more) firms, such as suppliers to buyers, allow firms to react in multiple fashions in the face of such economic shocks. The project examines such reactions when confronted with unforeseen contingencies.
The construction of a web of social relations surrounding them is one strategy that firms use to maintain their existence or even thrive in this unstable world. How firms build such a network, when and how these social relationships are mobilized is studied using data from Sweden, together with Oskar Nordström Skans (University of Uppsala). The relationships comprise family relations, past and present co-working relations, neighborhood relations, school or university-built relations, as well as relations built within a boardroom. Results show that social connections direct workers’ mobility across firms. In particular, we show that these “connected” moves help firms grow without deteriorating their productivity. The techniques used allow us to state these results in causal terms: “social connections cause firms’ growth”. Furthermore, even though connections per se appear to be positively sorted i.e. high-quality workers are connected to other high-quality workers who tend to be employed in high-paying firms, realized mobility tend to counteract this tendency by favoring movements to relatively low-paying firms, for all types of workers. Hence, social connections that could be thought of increasing inequality tend to decrease it when movements are taken account of.
When a board integrates a new member it creates an additional connection between two firms. And workers’ movements between the newly-connected firms increase markedly. Beyond workers’ movements, other firms’ outcomes are affected by board composition: having board members with family employed in a local bank office increases the firm’s propensity to finance using debt.
Another component of the project uses direct measures of firm-to-firm connections based on French Customs data to understand how firms react to shocks. Keeping in mind today’s circumstances, our goal is to understand how French firms adjust employment and wages when their clients “vanish”. Indeed, my previous research (together with Julien Martin and Isabelle Méjean), just published in the Journal of International Economics, has shown how (badly) diversified French exporters are. For the project, this three-authors team assesses how French firms react to the death of one of their clients. Given our previous research, we expect to see a sizable effect. The task is challenging because our data source does not measure directly clients’ death; it has to be inferred from the data. Various algorithms have been tested and we have assessed the impact of such a death on a French exporter total exports in order to understand the nature and magnitude of the shock. Indeed, our first results show that the (hypothetical) death translates into a loss of total exports and sales of approximately the size of exports to this buyer.
Another strategy to understand how firm-to-firm connections affect the economy is offered in a more structural project. This strand of the research proposes a totally new model of firm-to-firm connections based on random encounters between suppliers and buyers, a set of empirical moments from the data and their theoretical equivalents from the model, a full-fledge estimation of the models parameters, and a set of counterfactual predictions using the estimated parameters. In the model, a buyer can produce goods either with workers or with purchased inputs and chooses the cheapest. A producer is both a seller of goods and a buyer of inputs. Hence, a chain of exchanges “spontaneously” emerges throughout the world, model of value chains based on each firm’s comparative advantage and chance (in meeting with sellers), again something that we know clearly matters in the recent events, both in the trade skirmishes that took place in the last two years and in the pandemics. This roundabout process is shown to be stable in that it generates a General Equilibrium (GE). The different components in this economy are modelled (producers, wholesalers, retailers, services, households) in a unified framework allowing an analysis of global labor market reactions to shocks. For instance, workers may move from the goods producing sector to services, in reaction to shocks. Wages (of production workers, in particular) as well as prices react to international trade forces (tariffs, changes in transaction costs…). We illustrate the mechanisms at work by looking at the effects of Brexit on UK firms as well as those located in Europe. After Brexit, production gets repatriated within the UK but real wages of workers decrease. We also examine how the Hartz reforms that took place in Germany around 2003 affected the competitive position of German firms by increasing the supply of an unskilled workforce, resulting in a gain of market shares away from other European competitors. Important to stress, this model lends itself to a detailed study of the effects of the pandemics on trade, prices, and wages.
The model just described assumes the existence of a unique wage per skill whereas data show that wages for workers endowed with similar skills vary across firms in a systematic way. It also relies on competitive pricing of goods. However, when firms price their output to buyers at the second lowest price offered to the same buyer (i.e. Bertrand pricing), firms end up with positive profits (in contrast to the previous model). Hence, a phase of bargaining between workers and their employees is likely to take place. This setting is able to generate the observed firm-to-firm relations between a firm’s exports and imports and the wages a firm pay to both its skilled and unskilled production workers.
In the fourth project, Philippe Choné and the PI seek to further understand how the advent of recent economic forces affects the firm’s use of labor. Indeed, when workers’ skills must be sold as a “bundle”, i.e. when workers’ multidimensional skills cannot be sold separately, the labor market (general) equilibrium has particular features. In particular, workers with similar comparative advantage in the k potential skills match to firms with compatible technological preferences in these skills. Hence, comparative advantage in the skills-set matters when the absolute levels in skills endowments does not. In today’s world, some of the previously bundled skills are faced with the opening of a market for that exact skill, something we label “unbundling”. Unbundling may be due to a technological shock, such as the advent of Uber, that transforms a market (for taxi rides, in the case of Uber), a legislative change, again opening a market for a skill, a task, that could not be exchanged and priced (such as some types of business services in Germany before the Hartz reforms), or the extension of international trade.
When a board integrates a new member it creates an additional connection between two firms. And workers’ movements between the newly-connected firms increase markedly. Beyond workers’ movements, other firms’ outcomes are affected by board composition: having board members with family employed in a local bank office increases the firm’s propensity to finance using debt.
Another component of the project uses direct measures of firm-to-firm connections based on French Customs data to understand how firms react to shocks. Keeping in mind today’s circumstances, our goal is to understand how French firms adjust employment and wages when their clients “vanish”. Indeed, my previous research (together with Julien Martin and Isabelle Méjean), just published in the Journal of International Economics, has shown how (badly) diversified French exporters are. For the project, this three-authors team assesses how French firms react to the death of one of their clients. Given our previous research, we expect to see a sizable effect. The task is challenging because our data source does not measure directly clients’ death; it has to be inferred from the data. Various algorithms have been tested and we have assessed the impact of such a death on a French exporter total exports in order to understand the nature and magnitude of the shock. Indeed, our first results show that the (hypothetical) death translates into a loss of total exports and sales of approximately the size of exports to this buyer.
Another strategy to understand how firm-to-firm connections affect the economy is offered in a more structural project. This strand of the research proposes a totally new model of firm-to-firm connections based on random encounters between suppliers and buyers, a set of empirical moments from the data and their theoretical equivalents from the model, a full-fledge estimation of the models parameters, and a set of counterfactual predictions using the estimated parameters. In the model, a buyer can produce goods either with workers or with purchased inputs and chooses the cheapest. A producer is both a seller of goods and a buyer of inputs. Hence, a chain of exchanges “spontaneously” emerges throughout the world, model of value chains based on each firm’s comparative advantage and chance (in meeting with sellers), again something that we know clearly matters in the recent events, both in the trade skirmishes that took place in the last two years and in the pandemics. This roundabout process is shown to be stable in that it generates a General Equilibrium (GE). The different components in this economy are modelled (producers, wholesalers, retailers, services, households) in a unified framework allowing an analysis of global labor market reactions to shocks. For instance, workers may move from the goods producing sector to services, in reaction to shocks. Wages (of production workers, in particular) as well as prices react to international trade forces (tariffs, changes in transaction costs…). We illustrate the mechanisms at work by looking at the effects of Brexit on UK firms as well as those located in Europe. After Brexit, production gets repatriated within the UK but real wages of workers decrease. We also examine how the Hartz reforms that took place in Germany around 2003 affected the competitive position of German firms by increasing the supply of an unskilled workforce, resulting in a gain of market shares away from other European competitors. Important to stress, this model lends itself to a detailed study of the effects of the pandemics on trade, prices, and wages.
The model just described assumes the existence of a unique wage per skill whereas data show that wages for workers endowed with similar skills vary across firms in a systematic way. It also relies on competitive pricing of goods. However, when firms price their output to buyers at the second lowest price offered to the same buyer (i.e. Bertrand pricing), firms end up with positive profits (in contrast to the previous model). Hence, a phase of bargaining between workers and their employees is likely to take place. This setting is able to generate the observed firm-to-firm relations between a firm’s exports and imports and the wages a firm pay to both its skilled and unskilled production workers.
In the fourth project, Philippe Choné and the PI seek to further understand how the advent of recent economic forces affects the firm’s use of labor. Indeed, when workers’ skills must be sold as a “bundle”, i.e. when workers’ multidimensional skills cannot be sold separately, the labor market (general) equilibrium has particular features. In particular, workers with similar comparative advantage in the k potential skills match to firms with compatible technological preferences in these skills. Hence, comparative advantage in the skills-set matters when the absolute levels in skills endowments does not. In today’s world, some of the previously bundled skills are faced with the opening of a market for that exact skill, something we label “unbundling”. Unbundling may be due to a technological shock, such as the advent of Uber, that transforms a market (for taxi rides, in the case of Uber), a legislative change, again opening a market for a skill, a task, that could not be exchanged and priced (such as some types of business services in Germany before the Hartz reforms), or the extension of international trade.
In data that just became available, we directly measure vacancies as advertised by firms in order to test the idea that social connections are useful beyond filling vacancies. Therefore, such relations may help firms seize difficult-to-predict expansion opportunities and grow without posting vacancies in good times. In bad times, as is happening now, our analysis of buyer-supplier relations will provide the first direct measures of the international transmission of shocks on labor market outcomes. At all times, similar workers employed in different firms are payed differently. But why? Our model of firm-to-firm trade with Bertrand competition and bargaining will allow us to address this long-standing question in labor economics without relying on job-search modeling, with “real” firms (rather than firms seen as a collection of independent jobs). Beyond booms and busts, our analysis of the unbundling of "skills bundles" will show which firms and which workers benefit from this opening and, by contrast, which firms and workers are negatively affected. Because going from bundling to unbundling is new to the literature, we must also study facts, something we will do using in Swedish data sources on workers’ cognitive and non-cognitive skills.