Colombia

Organizational Responses to Product Cycles: Evidence from Auto Manufacturing

Product cycles entail the mass production of new – and often increasingly complex – products on a regular basis and are a fundamental feature of innovation, growth, and survival in many industries. How do firms manage such a highly dynamic production environment?

Product cycles are a fundamental feature of innovation, growth, and survival in many industries. Firms develop new generations of products to remain profitable when the production of current generations becomes imitable. These product cycles can be anywhere around 2 to 5 years depending on the industry. Our study focuses on the production of new models of automobiles, a prototypical example of product cycles. We combine granular administrative data from a leading global auto manufacturer with event study and discontinuity-based methods.

We find that the production of new models necessitates learning-by-doing, hence increasing the defects per vehicle substantially after the production change, which decrease to their prior level over a period of 2-3 weeks.

We analyze two types of discrete changes to the production process:

1. Model Shocks: Such changes result in a significant increase in the number of new parts that have to be produced and assembled, with resulting changes in the skills needed in production and the types of problems that arise on the assembly line
2. Volume Shocks:Such changes result in a substantial increase in the number of cars that have to be produced, but not in the complexity of what needs to be produced since the model variant does not also change at the same time. In this change, the plant needs to produce more of the same.

The plant has no discretion as to whether or when to implement these changes. Such decisions are made by the manufacturer; the plant is tasked only with executing production. We study the organizational responses put in place by a manufacturing plant as a response to such changes to bring down defects per vehicle and how they differ across the two types of shocks.

The plant’s model shock response and following results show how an increase in the complexity of the problems that need to be solved can lead to a reduction in organizational layers inconsistent with prior literature. On the other hand, the volume shock response shows how the organization controls negative productivity through a monotonic response: permanent increase in both employment and management layers, consistent with prior evidence from manufacturing in high-income countries.

We find that: 

1. Both volume and model shocks lead to a similar short-term reduction in productivity but the plant is able to bring these back down fairly quickly. 
2. In the model shock: 
   1. The number of managerial layers decreases on average by about one layer as a result of the reallocation of existing workers across layers, rather than by increasing or downsizing the workforce. This effect is stable and significant in the first three weeks after the shock.
   2. The reallocation of workers is found to be away from the middle layers and towards higher level positions. 
   3. The plant waits to back-fill the middle ranks positions. 
   4. Once working groups have learned the new tasks and the negative productivity impacts following the introduction of new model variants have been addressed, working groups again revert to the initial hierarchical structure. 
3. While volume changes lead to a similar spike in the incidence of defects per vehicle, the organizational response is very different. 
4. In the volume shock: 
   1. There is a sudden increase in the number of managerial layers within working groups going up by just under 1 layer. This increase is persistent for several weeks after the volume shock.
   2. The plant hires more entry-level workers and adds more layers to working groups, so that the distance – in terms of production layers – between front-line workers and supervisors further up the hierarchy increases.
   3. The overall plant employment increases by about 70-80 workers over the two months following the volume shock.
   4. The effect on employment is concentrated in frontline workers

While model shocks lead to workers being trained and promoted away from mid-level positions and towards higher level positions, making the organization both top and bottom heavy and reducing the distance between highly skilled workers and front line employees, volume shocks lead to an organization that is heavier in the middle and lower part.

Colombia

The Anatomy of Performance Monitoring

Can performance monitoring technology alone boost productivity gains? Does it affect hiring practices? Can managers implement other programs to support gains observed by monitoring technology?

Our partner QSR firm owns and directly operates more than 60 fast-food restaurants in Puerto Rico. Importantly, the main data systems utilized in the restaurants are also provided by the same partner. The drive-thru service was one of the brand’s most significant innovations in the industry. The drive-thru contributes approximately two thirds of the chain’s total sales. QSRs in the region, much like those across the globe, employ the Ford production line approach to fast food.

Beginning in 2019, the chain implemented in staggered fashion a drive-thru performance monitoring system in 51 Puerto Rico restaurants to track productivity at the primary point of sale. We know the exact week in which each store implemented the monitoring technology. Since each store implemented the technology at different times, we can measure the technology’s impact on different store performance indicators before and after the implementation.

We used data from three main sources. The first source is data on all the transactions of the stores that took place during our sample period. The second is employment and training data, and the third is waste and yield data for each store.

• While sales increased by nearly 5%, this rise was caveated by their short tenures. Impacts regressed to roughly half their initial magnitude within two months of the intervention.
• It was found that managers did respond to the availability of real-time data on bottlenecks by providing greater training inputs to workers at key workstations, particularly in the kitchen. However, only a subset of managers provided “refresher” training to counteract skill depreciation over time.
• Stores where managers utilized refresher training intensively prior to the technology implementation had more persistent gains in sales, suggesting that managers’ attention and responses to worker skill dynamics matter for productivity.
• Performance monitoring technology alone cannot ensure productivity gains and on-the-job human capital investment is critical to sustaining such efficacies.

Colombia

Managing Demand Shocks: Evidence from Quick Service Restaurants

Adapting to the impact of external shocks is an integral part of managing an organization– whether these are as drastic as the impact of the Covid-19 pandemic, or the less dramatic but still disruptive impact of technological advancements. While there is budding literature focused on the responsive adaptation of organizational hierarchy, this project focuses on peeking inside the “black box” of managerial adaptation by studying how the allocation decisions of managers change in the face of a large demand shock.

We partnered with a leading quick-service restaurant (QSR) firm that operates 76 restaurants located in 9 cities of Colombia. The firm averages a monthly sale of nearly 3 million units (items). Our partner implemented online ordering via a third-party delivery platform that brokered a deal to be the sole delivery service for its restaurant locations in Colombia.

In lieu of the deal, the delivery company scaled up their service (by hiring and training more drivers) in each city before making the app available to the QSR firm. The phased roll out of the service gave little control to store managers as to when the delivery channel was implemented, and hence little opportunity to prepare for the shock caused by increased demand. Since each store implemented the app delivery service at different times, we can measure the impact of the app on different store performance indicators before and after the implementation.

Empirically testing our hypotheses was made possible by the fact that almost all production processes, employee and manager functions, and capital are, by design, identical across the restaurants in our sample. We were also given access to records of how managers allocated workers to or changed their shifts, and what trainings were provided by managers to employees. On a personnel level, data included demographic characteristics (age, store, position, and gender) and the date of hiring and termination for each worker across all the restaurants.

1. There was an overall increase in demand due to the implementation of the delivery app, which was concentrated during peak hours (lunchtime and dinnertime). The increase was nearly immediate and persisted throughout an 18-week sample of post-implementation performance data. 
2. Managers met this increase in demand, and achieved an increase in performance without adding additional workers to their teams. This was accomplished with through 
   1. Changes in shift allocation: Managers rescheduled employees’ modal shifts to better meet times with peak demand, increased the number of shifts per employee, while also increasing the number of short notice reschedules. While it has been theorized that short notice schedules can cause an increase in turnover and absenteeism in the workforce, we surprisingly observed no significant changes to either at store level. 
   2. Increased training: Managers invested in training of their existing workforce, conducting more refresher trainings per store. Since workers must be trained before they can work different stations in a store, the additional trainings make it easier for managers to optimally allocate workers to meet demand. 
3. Gender-balanced stores did better in driving sales and matching productivity, in comparison to stores that had a gender mismatch between the management and workers. 
   1. The impact of women managers on performance increases with the share of women workers in the stores. The same holds true for men dominated teams led by managers who are men. However, productivity is considerably lower when the gender of the managers differs from that of the workers. 
   2. Stores with a gender mismatch – with either men-heavy management (for eg. a store with 20% men managers and 70% women crew) or women-heavy management (for eg. a store with 70% women management and 20% men crew) carried out fewer reschedules, and saw increased turnover and absenteeism in comparison to gender balanced stores.