nucleation

THEORY-852: Under Szeidl and Szucs's structural model and its identifying assumptions (an owner's ideology fixed across episodes; the government's online advertising budget not driven by the outlets' slant), both owner ideology and favor exchange drove the Hungarian portal Origo's reduced coverage of corruption in 2013–2016: favor exchange alone cut it by 29–44%, the new owner's ideology alone by 63% after 2016, and the two together by 74%, so they are substitutes; and a ban on targeted government advertising would have raised it by 41–79%

2 min read

Properties6
StatusActive
Version2
Tagssociety-and-governance, social-science
SourceLIT-1315 — Media Capture Through Favor Exchange
ExtendsTHEORY-860 — In Szeidl and Szucs's model of two outlets competing for readers and advertisers, where state-owned advertisers reward only the connected outlet's slant, with weight δ, every equilibrium has the slant gap between the connected and the unconnected outlet equal to their owners' ideal-point gap plus a favor-exchange term (βδ/NT)·Σ over government advertisers of masⱼWⱼ, up to measurement noise; readers' preference for slant cancels from the gap, so equal observed gaps can come from different mixes of ideology and exchange
Cited as “extends” byTHEORY-1217 — Both owner ideology and favor exchange drove the Hungarian portal Origo's reduced coverage of corruption in 2013–2016: favor exchange alone cut it by 29–44%, the new owner's ideology alone by 63% after 2016, and the two together by 74%, so they are substitutes; and a ban on targeted government advertising would have raised it by 41–79%

Graph View

Source

Szeidl and Szucs (2021), LIT-1315, read in NOTE-1039: Section 4.4 (identification), Table IV (estimates) and Table V (decomposition).

What was actually shown

The model of THEORY-860 was estimated by simulated method of moments on three online episodes (before Origo’s new editor, after it, after the new owner) and two print episodes (before and after the fallout). δ is identified from the co-movement of the connected outlet’s relative slant and government advertising. β is identified from periods in which the owner, and so by assumption the ideal point, is unchanged but the reward for slant changes: online, a rise in the government’s budget under Telekom; in print, the end of Simicska’s connection. Pooled estimates: δ = 441, γ = −69, β = 1.50, with bootstrapped intervals excluding zero. Letting Simicska’s ideal point change at the fallout gives a small, insignificant change.

Decomposition for Origo, against a world with no targeted advertising and Telekom as owner throughout: favor exchange cut corruption coverage by 29%, 44% and 42% in the three periods. The new owner’s ideology alone cut it by 63%, and the two together by 74%, less than the sum. Setting δ = 0 would have raised coverage by 41%, 79% and 42%.

What could have come out otherwise: β could have been zero, meaning slant did not respond to the reward with the owner fixed, so that all content favors were ideology. It was not, in any specification.

What this does not say

  • Not free of the identifying assumptions. If Simicska’s ideology changed at the fallout for reasons other than lost favors, or the government’s online budget responded to Telekom’s slant, β is misidentified. The authors give arguments against both, not tests that settle them.
  • Not about rewards other than advertising, or slant other than corruption coverage.
  • Not a prediction for a temporary ban. The authors note that a pause expected to end may be absorbed by a flexible exchange.
  • Not a dynamic account. Each episode is a separate equilibrium.
  • Not beyond Origo and Magyar Nemzet.

Backlinks