Hurricane Maria and Puerto Rico after 5 years

Now seems a useful time to reassess some of the statements made in the wake of the catastrophic disaster that struck Puerto Rico in September 2017. First, the deaths of Americans (contrary to Mr. Trump’s apparent assertion that they are not Americans) were much higher than some commentators had claimed. Second, the arguments that the economic policies taken in 2016 and 2017 (i.e. austerity measures associated with PROMESA) caused more deaths than Hurricane Maria are incorrect. Finally, the economic challenges that existed before the hurricane struck — including insufficient tax revenue — remain, even as the economy recovers.

Impact on excess mortality

First, let’s remember that some commentators argued in the aftermath of the hurricane that there were no more than 200-400 deaths. Even after they were educated in terms of what the data meant, some continued to give unrealistically low estimates.

From Sandberg et al. (July 2019) in Epidemiology:

Now compare with the comments of a hip shoot on Econbrowser, like this one from Stephen Cobbets on 05/31/2018:

The number of excess deaths in PR to the end of the year, those recorded by the Census Bureau, was 654. Most of these deaths occurred in the last ten days of September and the whole of October. While the blackouts there were exacerbated by state ownership of public relations facilities, it is likely that a large portion of the excess deaths occurred regardless, given the terrain and hurricane strength. Thus, there may have been 300-400 excess deaths regardless of what steps a person could take to repair the power supply. The rest can be attributed mainly to state ownership of the power facility.

I would note that excess deaths fell by half in December. Thus, the data suggests that the tornado hastened the deaths of the sick and dying, rather than killing them instantly. I project the number of excess deaths on a year horizon (during, say, October 1, 2018) to a total of 200-400. It’s still a notable number, but definitely not 4600. [emphasis added-MDC]

Kobets’ updated analysis (6/4), even with the updated data, concluded:

Thus, the year-end excess death toll of 1,400 may be treated as a fixed number in practice.

I think the word “firm” is an adjective that should be avoided in these situations. Below is a graph showing the selected estimates, from the April 2019 publication, which includes some of the earlier estimates.

Figure 1: Cumulative excess mortality from September 2017, for simple time dummies, OLS model (blue), modified population OLS model (green), modified population quantitative regression model (red), Milken Institute point estimate (black square) and 95% confidence interval (grey +) , Santos-Lozada, Howard’s letter (a striped triangle), Cruz-Cano and Mead (pink squares), Kopits (a greenish-blue triangle). Not pictured: Kopits estimate of 300-400 for October 2018. Source: author accounts, Milken Institute (2018), Santos-Lozada and Howard (2018), Cruz-Cano and Mead (2019), Kopits (2018).

Did austerity kill more than a hurricane?

What about the arguments that the austerity measures associated with PROMESA caused more excess deaths than the typhoon? [extensive argument here] . To assess this argument, first consider mortality data through February 2018.


Figure 2: Mortality rate per month (blue). Gray indicates the period of the sample; Orange shading indicates Hurricane Maria and the post-hurricane period; Dashed line in PROMESA legislation. Source: Santos-Lozada and Howard, 2017, June release of vital statistics data.

Second, consider now the counterfactual construction that does not include austerity measures before and after the implementation of PROMESA (legislation passed as of July 2016, censorship in effect as of October 2016). This was accomplished by estimating two equations: (1) a simple mean over the 2010-2015 period, and (2) a log-least-squares regression specification that includes population estimates (plus a dummy model for October 2014). 2016 seems an appropriate watershed for austerity given Brad Setser’s discussion of Puerto Rico’s finances. These specifications are discussed in this post. Present in Figure 3 the figures for the implied excess mortality.


Figure 3: Monthly excess mortality computed using 2010-15 averages (blue), population adjusted using 2010-15 sample (red), population is IMF interpolation cubed Global Economic Prospects database data. Gray indicates the period of the sample; Orange shading indicates Hurricane Maria and the post-hurricane period; Dashed line in PROMESA legislation. Source: Santos-Lozada and Howard, 2017, June release of vital statistics data, International Monetary Fund Fabulous April 2018 database, and author calculations.

Note in both cases that most of the pre-maria deviations are statistically significant at 10% ms. In other words, one cannot usually reject the null hypothesis that there was no increase in deaths due to austerity, pre-Maria.

Third, it is useful to consider the excess deaths from 2016M01-2017M08, and how they compare to the excess deaths from 2017M09-2018M02. If one assumes no population change from 2016 to 2017, one would get an estimate of the cumulative mortality (‘mean’ 10–16) in the red line, indicating a minimal impact of austerity.


Figure 4: Increase monthly cumulative mortality using the population adjustment specification (in blue) using the IMF’s cubic interpolation Global Economic Prospects Database data, average use 2010-15 (red). Orange shading indicates Hurricane Maria and the post-hurricane period; Dashed line in PROMESA legislation. Source: Santos-Lozada and Howard, 2017, June release of vital statistics data, International Monetary Fund Fabulous April 2018 database, and author calculations.

However, the most realistic assessment depends on adjusting the corresponding reality of the population. This leads to the blue line, called the “record”), which appears to validate the assumption that the excess deaths started before the hurricane made landfall. Interestingly, however, neither approach directly conflicts with the point that most excess deaths since 2016M01 are due to the impact of Hurricane Maria.

I conclude that using statistical analysis, the inference that the excess deaths due to austerity in the pre- and post-Maria period outweigh those from the aftermath of Hurricane Maria is a very fragile conclusion.

In support of this view, I also find, using Shermeyer (2018) power grid outage data, that excess deaths as calculated using a population adjustment match very closely with outage data, derived either from PERPA (public utilities) or from Suomi National Polar- Visible Infrared Imaging Radiometer Set (NPP VIIRS).


Figure 5: Monthly excess mortality using the population adjustment specification (black) using IMF cubic interpolation Global Economic Prospects Database data, and power outages as a percentage of the total, from PREPA (in pink) and from VIIRS as reported by Shermeyer (2018). Orange shading indicates Hurricane Maria and the post-hurricane period; Dashed line in PROMESA legislation. Source: Santos-Lozada and Howard, 2017, June release of vital statistics data, International Monetary Fund Fabulous April 2018 database, personal communication from Jacob Schirmayer and author accounts.

Regression over the period 2007M04-2018M02, the regression coefficient at cut-off is 826 using Puerto Rico Electric Power Authority (PREPA) data, and 950 using VIIRS data, both statistically significant using strong HAC errors, with adjusted R2 = 0.83 and 0.72, respectively. That means 639-690 excess deaths attributable to blackouts (and associated) in October, for example. (The cause of death in excess may be a direct power outage, or a communication outage and water service interruption associated with a power outage.)

Exactly how bad is the federal response to Maria in Puerto Rico?

Skipping the optics of paper towels, many reports (eg, DHS OIG) have accurate documentation of how poorly FEMA is performing. In addition, academic analyzes documented the differential response that occurred in the same period—much larger responses to hurricanes in Texas and Florida.

Economic Outlook

After a catastrophic failure to respond to the disasters, Puerto Rico’s economy recovered to pre-hurricane levels, yet the Covid pandemic was low. Figure 6 below shows the GDP and the Economic Activity Index.

Figure 6: GDP in million Ch2012$ (blue bar, logarithmic scale left), and Economic Activity Index (EDB-EAI) sa (blue logarithmic scale, right). The National Bureau of Economic Research identified peak-to-trough recession dates shaded in grey. source: BEA, Puerto Rico Economic Development Bank, and NBER.

The broader measure of activity – GDP – is only reported through 2020, while an indicator of economic activity – based on employment, electricity generation, gasoline consumption and cement sales – that runs through August 2022 measures only a portion of the economy. As shown, while there is a recovery, in economic activity (measured), it has eased in recent months. This pattern appears in civilian employment (albeit not sequentially from the chain of institutions).

Figure 7: Puerto Rico nonfarm employment (in blue), private nonfarm payroll employment (tan), civilian employment (green), all in 000, sa. The National Bureau of Economic Research identified peak-to-trough recession dates shaded in grey. source: BLS and NBER.

Federal aid linked to the pandemic is about to expire. These financial hurdles add to the central long-term challenges of an uncompetitive economy (due in part to shipping and other transport regulations) and huge government debt mired in restructuring (see CRS). See also Gregory Makoff and Brad Setser’s 2017 economic analysis.

Some of these challenges are intractable, while others can be addressed relatively easily. For example, a waiver of the Jones Act for Puerto Rico could in some estimates reduce prices by $1 billion, which is substantial for an economy measured at about $103 billion in 2020.

Related Posts

Leave a Reply

Your email address will not be published.