DEAS

In this paper we investigate the existence and the size of geographical differences in Italian students' mathematical competencies. We analyze a novel data set that combines the 2003 wave of the OECD Programme for International Student Assessment (PISA) with information about local economic conditions and school-level administrative data. We find there is significant positive correlation, across provinces, between mathematical literacy and school buildings maintenance and local employment probabilities. About 75% of the North-South differential in mathematical literacy is accounted for by resource differences, while geographical differences in school production functions account for the remaining fraction.

Realized wages are affected by investments and signals concerning productivity, which workers undertake on the basis of expectations on future job prospects. Thus, also the gender wage gap is likely to depend on workers' expectations. Using data on wage expectations of students and wage realizations of graduates from the same University (Bocconi, Italy), we show that the expected gender gap one year after graduation is consistent with the realized gap. There is instead a misperception later in the career because students expect the gap to be roughly constant while realizations indicate an increasing gap with experience. There is also evidence that the gender gap at the beginning of a career is particularly high in the most recent cohorts and lower in the previous ones. Finally, our results suggest that the careers of females are characterized by "glass ceilings" in particular at high skill levels, and by "sticky floors" at the opposite end of the skill spectrum.

This paper is an experimental analysis of the role played by workers' expectations in explaining the puzzling long-run persistence of observed discrimination against certain minorities in the labor market. The experiment provides some evidence supporting the theoretical prediction that unequal outcomes may emerge due to disadvantaged workers' wrong expectations of being discriminated against. However, this e¤ect is not long-lasting, since players learn the true state of nature in later stages of the experiment, failing to generate a Self-Con…rming Equilibrium driven by wrong beliefs. The strategy method provides additional evidence that expectations matter.

Any textbook states that high labor costs are negatively related to employment creation. We provide evidence that the two main components of labor costs (real wages and productivity) are not important correlates of net employment creation at the sector level in a sample of ten OECD countries in 1971-97. Our estimates show that the growth rates of real wages and TFP are statistically significant, but economically unimportant, correlates of employment growth. Instead, the bulk of the cross-section and time series variability of employment creation is accounted for by unobserved country-and sector-specific fixed effects. JEL Classification number: J21, J23, J24, J30

This paper presents empirical evidence from the PISA 2003 survey on the role of students' attitudes towards competition and cooperation in mathematical literacy achievement. While individual competitive attitudes are positively correlated with test scores, the reverse occurs when considering the aggregation of individual attitudes. Similarly, while individual cooperative attitudes exhibit a negative correlation

This paper is an experimental analysis of the role played by workers' expectations in explaining the puzzling long-run persistence of observed discrimination against certain minorities in the labor market. The experiment provides evidence supporting the theoretical prediction that unequal outcomes may emerge due to disadvantaged workers'expectations of being discriminated against that survive even after a real e¤ect, i.e. discriminatory tastes, has become negligible. A Self-Con…rming Equilibrium is observed in which wrong expectations of being discriminated against a¤ect subjects behavior and generate unequal outcomes that do not contradict their wrong beliefs. Even when learning takes place, disadvantaged subjects do not reach a situation of balanced outcomes. In other words, discrimination seems to have persistent e¤ect and expectations are likely to increase the probability that observed discrimination lags behind its causes.

Contributions to the public goods stem from selfish motivations (strategic cooperation) as well as other-regarding considerations, either related to opponents' behaviour (reciprocity), or unconditional. In this paper we provide strong experimental evidence that these components are not additive, and in particular that unconditional altruism is crowded out when strategic interaction driven by extrinsic motivations plays a role. We find that neutralizing strategic interaction in a repeated dictator games causes contributions to be not only significantly higher than in comparable standard linear public goods games, but also not characterized by the usual decreasing pattern. In contrast, when strategic interaction matters, participants behave less altruistically since the beginning of the game. Some of them start contributing a positive amount with the only purpose of increasing their individual payoffs, but such attempts are systematically frustrated causing the well-known decreasing pattern. We believe that guilt aversion is the most likely explanation of the different behaviour, since in the repeated dictator games the recipients' payoffs only depend on donors' generosity, while in the public goods game all the players are characterized by equal opportunities.

The paper investigates the effect of .99 price endings on consumer demand by means of a field experiment. Results tail behind other contributions showing how .99 endings can be ineffective, casting doubts on their widespread use among retailers. When the .99 ending price is removed an increase of sales emerges from descriptive statistics as well as a in multivariate framework in which only sales of the treated item are analyzed. However, such a counterintuitive effect does not survive in a diffs-in-diffs model in which also the daily sales of all the relevant substitutes are analyzed. Since a common shock at the time of the treatment does not emerge, the interpretation is that a different elasticity of demand drives the relative increase of sales during the treatment, when prices of the substitutes are on average higher. Once the different reactions to price changes are taken into account, the treated item does not display significantly higher sales as compared to its substitutes when the .99 ending price is removed. . I would like to thank Massimiliano Bratti for helpful suggestions and the participants at the ESA conference (Rome) for their comments. The usual disclaimer applies.

This paper aims at explaining the decoupling of official and perceived inflation emerged in the Euro area after the currency changeover in 2002. Such a phenomenon is puzzling since the surge of perceived inflation in the years after the introduction of the Euro is apparently not backed by any significant change in measured inflation. The question is therefore: what is wrong? Official statistics, perceptions, or a combination of the two? Or maybe something else? And why?

We investigate the emergence of discrimination in an experiment where individuals affiliated to different groups compete for a monetary prize, submitting independent bids to an auctioneer. The auctioneer receives perfect information about the bids (i.e. there is no statistical discrimination), and she has no monetary incentive to favour the members of her own group (the bidders are symmetric). We observe nonetheless some discrimination by auctioneers, who tend to assign the prize more frequently to a member of their own group when two or more players put forward the highest bid. Outgroup bidders react to this bias and reduce significantly their bids, causing an average decay of their earnings throughout the game, with cumulative effects that generate strongly unequal outcomes. Because the initial bias is costless, such mechanism can survive even in competitive market, providing a rationale for a well-known puzzle in the literature, i.e. the long-run persistence of discrimination.

In questo capitolo vengono utilizzati i dati italiani della wave 2003 dell'Indagine OCSE Programme for International Student Assessment (PISA) al fine di studiare le determinanti del divario territoriale che si registra tra nord e sud del paese. Grazie all'associazione dei dati di scuola alla provincia di localizzazione, si mettonoin luce le correlazioni con variabili relative allo stato di manutenzione degli edifici scolastici, alla spesa in conto capitale, al mercato del lavoro e ad alcune misure del capitale sociale. Tali correlazioni sono robuste alla analisi a livello di individui e di scuola, e per le diverse aree di competenza.

Il male non può essere estirpato. Nessuno è in grado di impedirne la crescita, in questo mondo. L'individuo può migliorare la sua condizione, forse, ma solo a spese degli altri. E ci saranno sempre re che si distingueranno l'uno dall'altro solo per il grado di crudeltà, ci saranno sempre baroni crudeli e debosciati, come ci sarà sempre la plebe stupida, la massa ignorante che ama i suoi oppressori e, paradossalmente, odia i suoi liberatori. Tutto questo si può spiegare con lo strano fenomeno secondo cui i servi e gli schiavi capiscono i loro padroni, anche i più crudeli, mentre non capiscono coloro che vorrebbero liberarli. Perché ogni schiavo sa immaginarsi al posto del proprio padrone, ma è raro quello che sa vedersi al posto del proprio liberatore. Questi sono gli esseri umani, Don Rumata. Così è il nostro mondo. Arkadi e Borsi Strugatzki, È difficile essere un dio, Marcos y Marcos 2005 (1964

In questo capitolo vengono utilizzati i dati italiani della wave 2003 dell'Indagine OCSE Programme for International Student Assessment (PISA) al fine di studiare l'effetto di alcune variabili individuali e di scuola sulle competenze acquisite dagli studenti in vari domini, controllando con effetti fissi provinciali l'impatto del fattore territoriale. I risultati mettono in luce che le correlazioni più forti sono legate a variabili di tipo familiare (prestigio occupazionale dei genitori, sia individualmente sia come media di scuola, possesso di libri e di risorse informatiche), così come sono connesse alla tipologia delle scuole frequentate. Minor impatto hanno invece le variabili relative al contesto scolastico, sia in termini di risorse che in termini di relazioni interne alla scuola. I dati segnalano altresì che le modalità di apprendimento sembrano premiare relazioni di tipo autoritativo da parte degli insegnanti, associato ad un controllo della disciplina in classe e ad un contesto di tipo competitivo per gli studenti.

Nelle classifiche redatte sulla base delle indagini internazionali sulle competenze degli studenti e degli adulti (come PISA, TIMSS, IALS) l'Italia si colloca sempre in posizioni piuttosto basse. Questa è una caratteristica comune ai paesi del bacino Mediterraneo, e potrebbe essere correlata al fatto che i miglioramenti nei livelli della scolarizzazione della popolazione siano storia relativamente recente. Al termine della seconda guerra mondiale l'Italia era ancora un paese in via di sviluppo, con più della metà della propria forza lavoro impiegata in agricoltura e una frazione simile di popolazione analfabeta. Cinquanta anni dopo, lo stesso paese si colloca tra quelli ad elevato livello di sviluppo, ma è ancora in posizione relativamente arretrata in termini di livelli di istruzione. Per questa ragione potremmo essere tentati di attribuire gli scarsi risultati degli studenti quindicenni in Italia alla presenza di un ambiente culturale carente: se questi studenti vivono in famiglie con scarsa istruzione, essi non ricevono sufficiente sostegno e pressioni familiari per conseguire dei buoni risultati a scuola. Tuttavia le cose sono probabilmente più complicate. Se osserviamo il livello mediano della performance delle regioni Italiane nei quattro domini di competenze testati in PISA (vedi Tabella 1), notiamo che la differenza mediana tra Nord e Sud raggiunge quasi una deviazione standard (per costruzione pari a 100 punti del test). Tabella 1: Performance mediane degli studenti nelle quattro aree di competenza dell'indagine PISA -Italia 2003 literacy matematica literacy letteraria conoscenze scientifiche problem solving NORD OVEST