Vertical Control in a Supply Chain and the Bullwhip Effect: An Experimental Approach

How to cite

Veri, Anna; Raff, Horst; Johnsen, Lennart. 2026. "Vertical Control in a Supply Chain and the Bullwhip Effect: An Experimental Approach." AEA RCT Registry. AEARCTR-0017474. https://doi.org/10.1257/rct.17474-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Anna Veri* , Horst Raff , Lennart Johnsen

* Primary investigator

Registry ID
AEARCTR-0017474
DOI
https://doi.org/10.1257/rct.17474-1.0
First registered
Last updated
Status
In development
Trial dates
2026-06-29 – 2027-06-28
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/17474

Abstract

We conduct a laboratory experiment to study how vertical control in supply chains affects the inci-dence and size of the bullwhip effect (BWE). The BWE refers to the phenomenon whereby small fluctuations in consumer demand translate into much larger swings in the quantities firms order from their suppliers. It can arise in decentralized supply chains comprised of separate, independently-owned firms, as well as in vertically integrated supply chains run by a singled firm. While vertical integration helps align pricing and inventory decisions, we hypothesize that it increases the incidence and the size of the BWE rather than reducing them.
In decentralized supply chains, upstream firms (manufacturers) and downstream firms (retailers) are independent decision-makers who must each form expectations about the others' choices before ob-serving them. This strategic uncertainty may lead participants to deviate from equilibrium behavior, amplifying order fluctuations beyond what demand shocks alone would predict. We therefore hy-pothesize that strategic uncertainty strengthens the BWE in decentralized supply chains.
To test these hypotheses in our experiment, participants take on the role of retailer and/or manufactur-er in a two-period supply chain game with random consumer demand. We vary the supply chain structure (decentralized versus vertically integrated) and the persistence of demand shocks. In some treatments we replace the human manufacturer with a computer to isolate the role played by strategic uncertainty.

Experimental design

The experiment uses a two-period supply chain game in which participants make decisions about sales quantity, order quantity, and, depending on the treatment, wholesale price, with the help of a calculator that displays all decision-relevant information, such as profit. The experiment is imple-mented in multiple sessions at the laboratory at Kiel University.
The experiment consists of six treatments that vary along three dimensions: the supply chain structure (vertically integrated vs. decentralized), the persistence of the demand shock, and, in decentralized supply chains, whether the manufacturer is played by a human or a computer.
The supply chain structure is either decentralized or vertically integrated. In decentralized treatments, participants are randomly assigned to one of two roles, retailer or manufacturer, and matched with another participant for the duration of the experiment; here, the manufacturer chooses the wholesale price, while the retailer chooses sales and order quantities. In vertically integrated treatments, a single participant makes all decisions for both stages of the supply chain, choosing sales and order quantities directly; no wholesale price exists in this setting.
The persistence of the demand shock takes a high or low value, determining how strongly a first-period demand shock carries over into the second period. Across all treatments, participants play 30 independent rounds, with the demand shock drawn randomly in each round.
In decentralized treatments, the manufacturer role is played either by a human participant or by a computer following a fixed rule; in the latter case, the human participant only takes the retailer role.
Of the 30 rounds, the first 10 are unpaid practice rounds, followed by 20 incentivized rounds. Pay-ments are based on cumulative profits earned in the incentivized rounds, converted to euros, plus a fixed participation fee. After the game, participants complete a questionnaire on their decision-making, a brief cognitive reflection test, and demographic questions.

Design details

Randomization method
random assignment of computer terminal and role played in the experiment at each session (either manufacturer or retailer) via random draw of number
Randomization unit
individual
Clusters
6 treatments (2 treatments with individuals representing either manufacturers or retailers, 2 treatments with individuals representing vertically integrated firms, 2 treatments with individuals representing retailers playing with a computer taking the role of the manufacturer).
Observations
240
Arms
30 individuals per role (retailer or manufacturer or vertically integrated firm) per treatment
Primary outcomes
participants' choices of sales quantities, order quantities, and wholesale prices

Cited by

No citation data available for this registration yet.

Keywords