(See also the NHH Bulletin summary.)
Digital platforms shape much of what we read, watch, listen to, and buy. Newspapers, streaming services, TV platforms, app stores, online shops, and booking sites all compete for users, but they do not compete in quite the same way as traditional brick-and-mortar firms. Once a digital service has produced or acquired content, serving one more user is often cheap ("it is free to copy bits and bytes"). This makes pricing, content choices, advertising, and platform strategy behave differently from markets where every extra customer comes with an extra cost.
The four papers summarised below study these issues using tools from industrial organization: the branch of economics concerned with how firms compete, how markets are structured, and how business strategies affect consumers. The first three papers focus on media platforms, such as newspapers and streaming services. These platforms usually offer a catalogue of content: subscribers do not buy one article, one song, or one episode at a time, but pay for access to a content catalogue. The fourth paper moves from media markets to algorithmic pricing, asking whether automated pricing bots behave in ways that look like tacit collusion.
A few economic concepts are useful before reading the summaries.
First, consumers differ. A left-leaning reader may prefer liberal newspaper, while a right-leaning reader prefers a conservative newspaper. A sports fan and a drama fan may value different streaming services. Economists call this horizontal differentiation: products are not simply better or worse, but better suited to different tastes. A standard way to model this is to imagine consumers spread along a line and products take a fixed position on the line; distance represents how far a product is from various consumers' ideal option. The further away the product is from a consumer, the less attractive it becomes to them.
Second, consumers can singlehome, multihome, or stay out of the market. A singlehoming consumer subscribes to one platform. A multihoming consumer subscribes to several. A consumer outside the market subscribes to none. This distinction matters because in many media markets, consumers choose to subscribe to more than one provider. A reader can subscribe to two newspapers; a household can have Netflix, Disney+, and HBO Max at the same time. Standard theoretical economics models of media markets often assume that everyone buys at least one product, but this can miss an important margin: firms may compete not only to steal subscribers from each other, but also to bring new consumers into the market.
Third, many media platforms are two-sided. They earn money from subscribers, but also from advertisers who want access to those subscribers’ attention. This changes pricing incentives. A reader with only one subscription may be more valuable to a newspaper’s advertisers than a reader who can also be reached elsewhere. As a result, the platform cares not only about how many subscribers it has, but also about what kind of subscribers they are.
Fourth, content can be exclusive or non-exclusive. Exclusive content is available only on one platform: a sports right, a columnist, a documentary, or a popular series. Non-exclusive content is available in several places: general news, weather reports, match reports, or widely distributed music. The usual intuition is that exclusivity is especially valuable because it helps a platform stand out. The papers below show that this intuition is sometimes right, but not always. If non-exclusive content attracts people who would otherwise not subscribe at all, it can also be strategically valuable.
The common theme across the four papers is therefore simple: in digital markets, the details of consumer choice matter. Whether consumers choose one platform, several platforms, or none can change how firms price, how content is distributed, how policy works, and how we should interpret algorithmic behaviour.
Paper 1: Multihoming and Market Expansion: Effects on Media Platforms’ Pricing and Content Creation Incentives
Eng: My first and third papers are closely connected. Media platforms such as newspapers, streaming services, and TV channels — where subscribers get access to a catalogue of content — compete both for subscribers and for advertisers. This paper studies competition strategy by platforms when some consumers subscribe to several platforms (multihomers), some to one (singlehomers), while others to none. Platforms can distribute content exclusively, such as a popular columnist, a special documentary, or a sports right only they offer, or non-exclusively, such as general news, weather updates, or match reports that consumers can find in several places. Standard economic theory suggests that exclusive content should be more valuable, because it helps a platform stand out. We show that this is not always true. When the market can expand (more consumers taking on their first subscription), better non-exclusive content can attract consumers who were previously outside the market altogether (for example, consumers who did not find any newspaper or streaming service worth paying for). At the same time, when both platforms become better, they can all raise prices, this makes some consumers who used to subscribe to both platforms drop their second subscription and keep only their favourite one. That matters for advertisement revenues to the platforms. A person with only one subscription can only be reached through that platform, while a person with several subscriptions can be reached elsewhere too. Advertisers therefore pay more for singlehoming consumers than for multihoming consumers. If this advertising difference is large enough, platforms may prefer to distribute non-exclusive content: it both brings in new subscribers from outside the market and turns some multihomers consumers into more valuable singlehoming consumers.
Nor: Min første og tredje artikkel er tett knyttet sammen. Medieplattformer som aviser, strømmetjenester og TV-kanaler — der abonnenter får tilgang til en katalog av innhold — konkurrerer både om abonnenter og annonsører. Denne artikkelen studerer plattformenes konkurransestrategier når noen forbrukere abonnerer på flere plattformer (multihoming), noen bare på én (singlehoming), mens andre ikke abonnerer på noen. Plattformene kan tilby eksklusivt innhold, som en populær kommentator, en spesialdokumentar eller en sportsrettighet bare de har, eller ikke-eksklusivt innhold, som generelle nyheter, værmeldinger eller kampresultater som finnes flere steder. Standard økonomisk teori tilsier at eksklusivt innhold bør være mest verdifullt, fordi det gjør at en plattform skiller seg ut. Vi viser at dette ikke alltid stemmer. Når markedet kan vokse — ved at flere velger å ta sitt første abonnement — kan mer ikke-eksklusivt innhold tiltrekke forbrukere som tidligere ikke abonnerte på noe, for eksempel fordi de ikke syntes noen avis eller strømmetjeneste var verdt å betale for. Samtidig kan begge plattformene sette opp prisene litt når de blir bedre. Da vil noen som tidligere abonnerte på begge, droppe sitt andre abonnement og beholde bare favoritten. Dette er viktig for annonseinntektene. En person med bare ett abonnement kan kun nås gjennom én plattform, mens en person med flere abonnementer også kan nås andre steder. Annonsører betaler derfor mer for forbrukere med ett abonnement enn for forbrukere med flere. Hvis denne forskjellen er stor nok, kan plattformene foretrekke ikke-eksklusivt innhold: Det trekker inn nye abonnenter og gjør noen multihomere om til mer verdifulle singlehomere.
Paper 2: Channel Coordination on Exclusive vs. Non-Exclusive Content Under Endogenous Consumer Multihoming
Eng: This paper asks why in some markets (e.g. music streaming) some premium content is available almost everywhere, while in other content is locked to one platform (e.g. video streaming). In order to gain insight into what drives market outcomes, we study the case when new content is available, and a content provider and competing platforms deal to distribute it. The answer depends on how much exclusive content the platforms already have. When platforms are initially quite similar, as Spotify and Apple Music largely are, consumers have little reason to subscribe to both. They normally choose one service (singlehome). In such markets, both the platforms and a content provider (for example, a music label, film studio, or sports rights holder) prefers to sell its content to both platforms and charge a fee per subscriber. Because each extra subscriber now comes with a payment to the content owner, the platforms have less incentive to cut prices aggressively. The result is non-exclusive content distribution. When platforms already have many exclusives, as Netflix, HBO Max, or Disney+ do, some consumers are willing to subscribe to several services because each offers something different. In that setting, the logic changes. If both platforms get the same new content, it adds little value to consumers who already subscribe to both. The content owner therefore earns more by making the content available on only one platform. A revenue-sharing contract is better suited for this, because it lets the content owner share in the extra value created by exclusivity without pushing up the platform’s subscription price in the same way as a per-subscriber fee. The chain of logic is therefore: low initial exclusivity leads to consumer singlehoming, which favours per-subscriber fees and non-exclusive distribution; high initial exclusivity leads to multihoming, which favours revenue sharing and exclusive distribution.
Nor: Denne artikkelen spør hvorfor premiuminnhold i noen markeder, som musikkstrømming, er tilgjengelig nesten overalt, mens i andre markeder er innhold er låst til én plattform, som i videostrømming. For å forstå hva som driver disse markedsutfallene, studerer vi en situasjon der nytt innhold blir tilgjengelig, der nytt innhold blir tilgjengelig, og en innholdsleverandør tilbyr konkurrerende plattformer tilgang til å distribuere det. Svaret avhenger av hvor mye eksklusivt innhold plattformene allerede har. Når plattformene i utgangspunktet er ganske like, som Spotify og Apple Music stort sett er, har forbrukerne liten grunn til å abonnere på begge. De velger vanligvis én tjeneste; de singlehomer. I sånne markeder foretrekker både plattformene og innholdsleverandøren (for eksempel et plateselskap, filmstudio eller en rettighetshaver til sport) at innholdet selges til begge plattformer mot en betaling per abonnent. Siden hver ekstra abonnent da innebærer en betaling til innholdsleverandøren, har plattformene mindre insentiv til å kutte prisene aggressivt. Resultatet blir ikke-eksklusiv distribusjon. Når plattformene allerede har mye eksklusivt innhold, som Netflix, HBO Max og Disney+ har, er noen forbrukere villige til å abonnere på flere tjenester fordi hver av dem tilbyr noe forskjellig. Da endres logikken. Hvis begge plattformer får det samme, nye innholdet, gir det liten ekstra verdi for forbrukere som allerede abonnerer på begge. Innholdsleverandøren tjener derfor mer på å gjøre innholdet tilgjengelig på bare én plattform. En inntektsdelingskontrakt egner seg bedre til dette, fordi den lar innholdsleverandøren ta del i merverdien som skapes av eksklusivitet, uten å presse opp abonnementsprisen på samme måte som en betaling per abonnent. Logikken er dermed: lav eksisterende eksklusivitet leder til singlehoming, som gjør betaling per abonnent og ikke-eksklusiv distribusjon mer attraktivt; høy eksisterende eksklusivitet leder til multihoming, som gjør inntektsdeling og eksklusiv distribusjon mer attraktivt.
Paper 3: Media Policy and Newspaper Competition with Consumer Multihoming in an Uncovered Market
Eng: This paper asks whether common media policies actually make more people subscribe to newspapers. Governments often support journalism through lower VAT rates and direct subsidies, hoping to increase circulation and media pluralism among consumers. The paper studies a market where some readers subscribe to one newspaper (singlehomers), some subscribe to several (multiple), and some subscribe to none. The main point is that policy effects depend on whether newspapers are print or digital. For printed newspapers, each extra subscriber means extra printing and distribution costs. In that setting, reducing VAT works much as intended: newspapers lower prices, and more people subscribe. Digital newspapers are different. Serving one more reader costs almost nothing, while that reader may bring advertising revenue and subsidies. In such markets, a VAT reduction can have the opposite effect: because newspapers keep more of each subscription payment, they may raise prices, which can reduce circulation. Direct per-subscriber subsidies are more reliable, because they make each additional subscriber more valuable and therefore give newspapers a reason to lower prices and attract more subscribers. The paper also shows that media support affects content choices. Both lower VAT and direct subsidies make exclusive content more attractive for newspapers. Exclusive content can increase circulation and encourage some readers to take a second subscription, but it also means that readers of different newspapers are exposed to less common content. The policy trade-off is therefore straightforward: support measures may increase subscriptions, but they can also push newspapers towards more separate content universes.
Nor: Denne artikkelen spør om vanlige mediepolitiske virkemidler faktisk får flere til å abonnere på aviser. Myndigheter støtter gjerne journalistikk gjennom lavere merverdiavgift og direkte subsidier, med mål om å øke avisenes rekkevidde og styrke mediemangfoldet blant leserne. Artikkelen studerer et marked der noen lesere abonnerer på én avis (singlehomers), noen abonnerer på flere (multihomers), mens andre ikke abonnerer på noen. Hovedpoenget er at effekten av politikken avhenger av om avisene er trykte eller digitale. For trykte aviser innebærer hver ekstra abonnent økte kostnader til trykking og distribusjon. I slike markeder virker lavere merverdiavgift omtrent som ønsket: avisene setter ned prisene, og flere abonnerer. Digitale aviser er annerledes. Det koster nesten ingenting å betjene én ekstra leser, samtidig som leseren kan gi annonseinntekter og subsidier. I slike markeder kan lavere merverdiavgift få motsatt effekt: Fordi avisene beholder en større del av abonnementsinntektene, kan de velge å øke prisene, noe som kan redusere sirkulasjonen. Direkte subsidier per abonnent er mer treffsikre, fordi de gjør hver ekstra abonnent mer verdifull og dermed gir avisene insentiv til å senke prisene og tiltrekke flere abonnenter. Artikkelen viser også at mediestøtte påvirker innholdsvalg. Både lavere merverdiavgift og direkte subsidier gjør eksklusivt innhold mer attraktivt for avisene. Eksklusivt innhold kan øke sirkulasjonen og få noen lesere til å ta et andre abonnement, men det betyr også at lesere av ulike aviser i mindre grad eksponeres for felles innhold med mindre de har flere abonnementer. Det mediepolitiske dilemmaet er derfor tydelig: Støtteordninger kan øke antallet abonnementer, men kan også trekke avisene i retning av mer adskilte innholdsuniverser.
Paper 4: Do Capacity Constraint Bots Collude?
Eng: This paper is about pricing bots — algorithms that set prices automatically, for example in online shops. A concern in competition policy is that such bots may learn to keep prices artificially high, even when firms have not made any illegal agreement. The idea is that the bots may learn a kind of “punishment strategy”: If one seller lowers its price to attract customers, the rival bot responds with low prices for a while, making the price cut unprofitable. Previous research has shown that pricing bots can end up with high prices, but it often assumes that firms can sell unlimited quantities. That is unrealistic in many markets. Online shops can run out of stock, hotels can run out of rooms, airlines can run out of seats, and delivery services can run out of available time slots. We therefore study what happens when pricing bots compete while also facing capacity limits. The result is twofold. The bots often still set prices above the competitive level, but their behaviour does not look like clear collusion. To test this, we force one bot to lower its price and observe how the other responds. If the other bot truly “punished” the price cut, that would be a sign of problematic pricing. We do not find this. Instead, the bots’ behaviour depends strongly on how they are programmed to handle situations where more customers want to buy than there are goods, seats, or slots available. If they are simply given a capacity limit, they often set higher and more stable prices. If they are instructed to serve the customers with the highest willingness to pay first, prices often fall. If they serve customers more randomly, prices become much more unstable. The broader lesson is that high average prices are not enough to conclude that pricing bots have learned to collude. What matters is the strategy behind the prices.
Nor: Denne artikkelen handler om prisroboter — algoritmer som setter priser automatisk, for eksempel i nettbutikker. En bekymring i konkurransepolitikken er at slike roboter kan lære å holde prisene kunstig høye, selv uten at bedriftene har avtalt noe ulovlig. Tanken er at robotene kan lære seg en slags «straffestrategi»: Hvis én selger setter ned prisen for å kapre kunder, svarer den andre roboten med lave priser en stund, slik at priskuttet ikke lønner seg. Tidligere forskning har vist at prisroboter kan ende opp med høye priser, men den antar ofte at bedriftene kan selge ubegrenset mye. Det er ikke realistisk i mange markeder. Nettbutikker kan gå tom for varer, hoteller kan gå tom for rom, flyselskaper kan gå tom for seter, og leveringstjenester kan gå tom for ledige tidsluker. Vi undersøker derfor hva som skjer når prisroboter konkurrerer, men samtidig har begrenset kapasitet. Resultatet er todelt. Robotene setter ofte fortsatt priser over konkurransenivået, men atferden deres ser ikke ut som tydelig samarbeid. For å teste dette tvinger vi én robot til å sette ned prisen og ser hvordan den andre reagerer. Hvis den andre roboten virkelig «straffet» priskuttet, ville det vært et tegn på problematisk prissetting. Det finner vi ikke. I stedet ser robotenes oppførsel ut til å avhenge sterkt av hvordan de er programmert til å håndtere situasjoner der flere vil kjøpe enn det finnes varer eller plasser. Hvis de bare får en kapasitetsgrense, setter de ofte høyere og mer stabile priser. Hvis de instrueres til å betjene de mest betalingsvillige kundene først, faller prisene ofte. Hvis de betjener kunder mer tilfeldig, blir prisene langt mer ustabile. Den bredere lærdommen er at høye gjennomsnittspriser ikke er nok til å konkludere med at prisroboter har lært å samarbeide. Det avgjørende er strategien bak prisene.