The third chapter studies the welfare consequences of the destructive creation (bank
branches replaced by internet banking) of the US commercial banking
market following the Great Recession of 2009. Using a structural model,
we find that the cleansing effect (closure of unproductive bank branches)
of the recession increases the units of internet banking by about 56\% in 2016, compared to the case where the cleansing effect is absent. The share of internet banking in the retail service market is increased from 48\% to 60\% and the price of internet banking service is decreased by a factor of 16 by the cleansing effect of the Great Recession.
The two changes lowers the price of retail banking services in 2016 by 37\%: 53\% of the price reduction is attributable
to the replacement
of branches by internet banking and 47\% is attributable to the reduction of the price of internet banking. However, this cleansing effect also
results in a 2.5\% decrease in small business services in small cities.
These findings suggest that the cleansing effect of the recession benefits
retail consumers. However, small business lending may suffer.
The fourth chapter evaluates how information technology (IT) improvements contribute to the decline of small business lending in the US commercial banking market from 2002 to 2017. This paper estimates a general equilibrium dynamic model with banks that differ in size and choose the level of the transaction (hard information intensive) and relationship (soft information intensive) lending. The model shows that banks’ costs of evaluating borrowers’ hard information declined over this period by 46\%, and small business loans fell by 7\% (12\% in the data). This paper finds that banks’ higher reliance on IT to issue transaction loans is responsible for 37\% of the decline in the data, and the consolidation caused by IT improvements caused 22\% of the decline. Contrary to previous findings, this paper finds that when general equilibrium is considered, policy protecting small banks cannot increase small business lending.
We examine the relation between high frequency quotation and the behavior of stock prices between 2009 and 2011 for the full cross section of securities in the US. On average, higher quotation activity is associated with price series that more closely resemble a random walk, and significantly lower cost of trading. We also explore market resiliency during periods of exceptionally high low-latency trading: large liquidity drawdowns in which, within the same millisecond, trading algorithms systematically sweep large volume across multiple trading venues. Although such large drawdowns incur trading costs, they do not appear to degrade the price formation process or increase the subsequent cost of trading. In an out-of-sample analysis, we investigate an exogenous technological change to the trading environment on the Tokyo Stock Exchange that dramatically reduces latency and allows co-location of servers. This shock also results in prices more closely resembling a random walk and a sharp decline in the cost of trading.