Greetings everyone! Dimitris Papanikolaou and I just finished a paper looking at some of the supply side effects of Covid-19 using data on workers and firms. We use SafeGraph data to validate a measure of the extent to which workers can work from home. We will be posting all of these measures soon, but we already have a high-frequency indicator based on stock prices which tracks news about the lockdowns posted. Paper and data are available here: “Working Remotely and the Supply-side Impact of Covid-19” data page
@Lawrence_Schmidt_MIT_Sloan Really robust paper with implications on the disruptions of industry-specific supply side. I especially liked your argument of how certain demographic groups (e.g. young women with children without a college degree) are affected. I have some trouble understanding some of your broad outcomes. Would you say your forecasts are more pessimistic than analysts’ forecasts? Are these two even comparable?
@Lawrence_Schmidt_MIT_Sloan I’m surprised by the -38.9% coefficient between Covid-19 work exposure and stock returns reported in Table 4. I’m assuming the -38.9 means that if the Covid-19 exposure went from 0.1 to 0.9, then the expected stock return would drop by 0.8*38.9 = 31.12%. I’m guessing the results are robust, although, as is typical in financial research, the R2 is low. Still fascinating. On the critical vs. non-critical in Panel B of Table 4, I’m also surprised that the difference is only -25.6 vs. -8.0 (17.6%). Seems like it would be bigger. In any event, I enjoyed the paper.
Hi guys, sorry for being a little slow to get back to you (haven’t been monitoring the slack). @Ruowei_Yang_UM_Baltimore: we aren’t making specific forecasts of our own, but the stock price reactions are quite large (at least relative to revenue changes). This could make sense due to a combination of two factors: 1) earnings could be more sensitive than revenues to the supply-side disruptions (e.g., firms cannot cut many of their sources of costs to absorb the shock, may have to incur higher ones in order to stay open, or face downward pressure on profit margins) and 2) investors’ required compensation for bearing the risk of holding stocks which are likely to lose a lot in value if we get bad news about COVID likely increased a lot. We hope to drill down more into these issues in future versions.
@Thomas_Young_Econometric_Studios_Utah_Legislature, yes your interpretation of the coefficient is correct. For a one month change, those magnitudes are extremely high, though we hope that some of these employment effects will be more transitory in nature. Low R2 isn’t terribly surprising in this context in part, in part because our measure is based on surveys which don’t always have large numbers of responses for some industries (so there is potential for measurement error). Often we find higher explanatory power if we use more aggregated industry classifications. Regarding the difference between critical and non-critical industries, I guess this is a question of your priors, but in my view an 18 pp gap over the length of a quarter is quite large (e.g, if you were to compare with other recessions), and the long run discrepancies are also quite notable.