HERDING IN EQUITY CROWDFUNDING UNDER VARYING BEHAVIOURAL CUES: EVIDENCE FROM THE ITALIAN MARKET

THIS THESIS INVESTIGATES HOW HERDING DYNAMICS, DEFINED AS INVESTORS CONDITIONING THEIR DECISIONS ON THE OBSERVED ACTIONS OF OTHERS, UNFOLD IN THE ITALIAN EQUITY CROWDFUNDING MARKET. BUILDING ON SIGNALLING THEORY AND THEORETICAL MODELS OF INFORMATION CASCADES AND HERDING, THE ANALYSIS FOCUSES ON SOCIAL SIGNALS GENERATED BY PRIOR INVESTMENT ACTIVITY. USING HIGH-FREQUENCY PLEDGE-LEVEL DATA ON APPROXIMATELY 26,000 INVESTMENTS ACROSS 226 EQUITY CROWDFUNDING CAMPAIGNS HOSTED ON A LEADING ITALIAN PLATFORM BETWEEN 2015 AND 2025, THE THESIS EXAMINES HOW THE STRENGTH AND FORM OF SEQUENTIAL DEPENDENCE VARY ACROSS INFORMATIONAL SETTINGS THAT PROXY DISTINCT BEHAVIOURAL CUES. FOUR RESULTS EMERGE. FIRST, APPARENT HERDING IN PARSIMONIOUS SPECIFICATIONS VANISHES ONCE CAMPAIGN FIXED EFFECTS AND CALENDAR-DAY SHOCKS ARE CONTROLLED FOR. SECOND, UNUSUALLY LARGE PLEDGES ARE TYPICALLY FOLLOWED BY SMALLER ONES, PRODUCING SHORT-RUN OSCILLATIONS RATHER THAN SELF-REINFORCING ESCALATION IN PLEDGE SIZE. THIRD, AFTER CONTROLLING FOR ENDOGENEITY IN PLEDGE TIMING, THE ELAPSED TIME BETWEEN CONTRIBUTIONS HAS NO MEANINGFUL CAUSAL EFFECT ON SUBSEQUENT PLEDGE SIZE, IMPLYING THAT TEMPORAL CLUSTERING PRIMARILY REFLECTS SYNCHRONISED ATTENTION RATHER THAN CONTAGION. FOURTH, CAMPAIGN-STAGE PROGRESSION, MEDIA COVERAGE, AND INVESTOR Q&A ACTIVITY SHIFT AVERAGE PLEDGE LEVELS BUT DO NOT GENERATE PERSISTENT IMITATION IN PLEDGE MAGNITUDES. OVERALL, THE FINDINGS PORTRAY HERDING IN EQUITY CROWDFUNDING AS BOUNDED AND CONTEXT-DEPENDENT, THEREBY REFINING THE CURRENT UNDERSTANDING OF SOCIAL LEARNING IN FINTECH-MEDIATED ENTREPRENEURIAL FINANCE.