Working Paper: CEPR ID: DP4917
Authors: Marc Ivaldi; Szabolcs Lőrincz
Abstract: The paper defines, implements and compares two empirical tests of relevant markets. While the traditional SSNIP test compares an initial industry equilibrium to an out-of-equilibrium situation, the FERM test, our contribution, compares the same initial equilibrium to an other equilibrium outcome. Hence, it is more in line with the behavioural assumptions of the underlying model of industry equilibrium and this can have significant consequences. We define these concepts formally and apply them to the industry of computer servers by estimating a model on a large dataset. We find several smaller relevant markets in the low-end segment of servers.
Keywords: computer servers; differentiated products; relevant market tests
JEL Codes: C33; L41; L63
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| traditional SSNIP test (L49) | misleading conclusions about market power and relevant market sizes (L11) |
| hypothetical cartel could profitably raise average prices (D43) | SSNIP test findings (H29) |
| FERM test (C93) | more consistent with underlying economic theory (E19) |
| FERM test (C93) | captures dynamics of competition more accurately (D43) |
| high differentiation among products (L15) | constrains substitution opportunities (D10) |
| proper application of SSNIP and FERM tests (L11) | more fragmented market structure (L19) |