Financial decision-making and the role of financial technology

Abstract

The overarching theme of this thesis is: "Financial Decision-Making and the role of Financial Technology”. The thesis uses new forms of data to test existing assumptions in the areas of behavioural and household finance. It improves our understanding of how we can use technology to debias financial decision-making. The thesis is based on three academic journal articles, each answering an individual research question. The first paper: “Transparency and Investment Decisions” analyses a novel data set from a social trading platform called wikifolio.com. The users of this platform create equity portfolios and by showing their complete trading history as well as all current holdings, try to convince other users to invest money into their own portfolio. Using proprietary data from this social trading platform, I follow a difference-in differences approach, analysing trading behaviour within and between portfolios as they become publicly visible. Portfolio managers on this platform go through three different stages of trading. Each phase is characterised by an increased level of transparency towards potential followers of the portfolio manager. In the last phase, the investable phase, all past trades and all current holdings of the investor are visible to the public. I find that the tendency of investors to forgo loss realisation in favour of gain realisation, known as the disposition effect, is considerably reduced in a transparent trading environment, i.e., when investors move through the phases. I find strong presence of the disposition effect in my sample, consistent with prior literature, but, I also show that the effect diminishes by about 36% when holdings and trades are publicly revealed. My findings are robust to survivorship bias, learning, and level of assets under management. This significant reduction in the disposition effect highlights the role of saliency and transparency of information in investing. The second and third chapter use data from one of the UK's largest banking transaction aggregation app, Money Dashboard (MDB). The data set is comprised of approximately 400 million banking observations from more than 300,000 individuals based in the UK. The transactions are automatically tagged by MDB with the type of expense (groceries, mortgage payments, dining and drinking, etc.). In addition, user login data and demographic data such as gender, age, income and address are stored. Since MDB collects historic data once users sign up, banking transactions are available for up to one year before an individual joined this service. This allows the present research to compare user behaviour before and after joining MDB. One of the functions provided to users is a budgeting tool. Users can set budgets for 270 categories and MDB automatically tracks and allocates their expenses to the individual budget. The second chapter titled: "The Influence of Budgeting on Spending” uses data from the budgeting tool and analyses the effect of spending goals on spending behaviour. This study is the first in the field to examine the influence of budgets on consumer spending using real world budgeting and spending data. The data reveals that budget compliance is weak: on average, consumers spend 21% more than they budgeted. On the other hand, the data also suggests that budgets do influence spending. Average spending in the month after budget creation is 11% lower than pre-budget spending. The effect is stronger for those users who log in frequently. The effect is surprisingly persistent; six months after setting a budget, spending is still significantly lower than pre-budget levels. These within-subject results are further supported by a quasi-experimental propensity score matching procedure that compares the spending of consumers with budgets with consumers without budgets. Taken together, these findings provide evidence that the influence of budgets on consumer spending is economically meaningful when actively tracked. The third chapter, "Payday - Dashboard Effect on Consumption Smoothing", studies the extent to which individuals are smoothing their consumption across the month. Economic theory suggests that the timing of a predictable future income should not influence the way individuals spend on that particular day. In contrast to this, I find that individuals spend 53% more on discretionary items three days after payday than they do during the rest of the month. This effect is commonly known as the payday effect. It seems that additional saliency of historic and present transactions helps individuals to smooth their consumption. Average spending for discretionary items was on average 65% higher around payday before individuals joined money dashboard and only 50% higher than usual after they joined. Once users frequently check their financial situation, they significantly change their spending behaviour and show smoother consumption patterns. This suggests that increased attention on spending patterns supports consumption smoothing. I find that if users log in once per month they only spend 28% more on discretionary items three days after payday and only 21% more if they log in at least six times in the respective month. However, only limited learning seems to take place, with users reverting to old habits of increased spending activity around the payday once they stop logging in. These findings suggest that once individuals pay less attention, they start spending more money on discretionary items. These findings contribute to the public debate on whether financial technology can help improve consumer financial decision-making. It also contributes to the neo-classical economic theory and suggests that individuals might require additional support structures to ensure smoothed consumption patterns. Overall, this thesis contributes to the behavioural and household finance literature by providing large-scale evidence for behaviour change after introducing new forms of financial technology. It explores three main areas of financial decision making: investing, budgeting and spending. While biased decision-making may persist, the thesis suggests that financial technology may be helpful in improving individual financial decisions. It appears that especially the framing and saliency of financial information supports behaviour change, but only if individuals pay attention. This thesis also informs policy makers and financial technology companies by showing that increased saliency and transparency of financial information helps to debias decision-making. Policy makers might want to use these insights and make sure that financial information from financial institutes has to be provided in a more salient way by aggregating the required information disclosed to clients. In addition to this, debt management programs and similar advice agencies might want to use the presented insights and ensure that their clients use financial technology to better understand their past and present financial decisions. The present results in the thesis suggest that financial technology, if used sensibly, might help individuals improving their investing, budgeting and spending behaviour.