Khora is processing your request...

Photo by Morning Brew on Unsplash


“Yet knowing how way leads on to way, I doubted if I should ever come back… Two roads diverged in a wood, and I— I took the one less traveled by, And that has made all the difference”. Robert Frost (1916)

1 – Introduction

During the initial stages of the COVID-19 crisis, ABEXUS Analytics published the first economic impact report on the effects of such devastating force on the main indicators of the Puerto Rican economy. Even though we pointed out that forecasts (including ours) should be taken with caution, recent estimates by other third-party entities have somewhat validated our initial assessment. For such, we feel humbled that our margin of error has been kept at bay. Nonetheless, the real judge will be the aggregated performance of economic actors and the “real” numbers that should be published in the following years.

Yes, as your read, YEARS! That is the amount of time that is required to get “accurate” or official data on the Puerto Rican economy. Even during the crisis, Puerto Rico’s official data on consumer expenditures dates from 2002. YES, almost 20 years! For a survey which is performed in the Continental U.S. on an annual basis. Even if some may consider such issue another statistical exercise, well, the fact is that our current inflation measure is based on this survey, moreover, even Child Support Guidelines all across the U.S. are based on such data. Hence, how much a child will have on its plate to eat, and how such food is deflated, depends on these data points.

Could you imagine driving your car with a gas gauge that is 20 years behind? Such, scenario has been sustained for too many years now. The experience citizens have had with the COVID crisis and its data inconsistencies is what a data analysist experiences on a daily basis in Puerto Rico. But not all the blame is on the public side, sometimes getting a reliable dataset from a private company is even harder than accessing public data. As a matter of fact, in a survey conducted by ABEXUS Analytics during May 2020, -in a scale from 1 to 10-, one out of every three participants informed no trust on official Government statistics related to COVID-19.

However, as a data analytics company, we know that government data is not the only source of “accurate” data on the economy. In fact, businesses and consumers generate an ocean of data points which are usually laying around and waiting to be analyzed. Keep in mind, that economic figures are also a byproduct of the technology that was available in a particular point in time. For instance, in the past we had to make surveys for almost every datapoint, and even today they remain valid (we even use them in our day-to-day operations), but it will be a shame to ignore the gigantic datasets that are produced directly from economic activity, which is sometimes the universe of datapoints, rather than a sample.

2 – Non-Traditional Data, for an Uncommon Collective

With this in mind, and with no intention to step into the realm of subject-matter experts, ABEXUS Analytics is publishing a series of non-traditional economic measures to gauge the performance of the economy during the COVID-19 crisis. As the title of this note suggest, we will focus on a set of alternative data sources that we have acquired or which are published by private companies.

As traditionally-trained economists ourselves- we started by trying to get the most accurate, up-to date and reliable figures on the economy of Puerto Rico. The results from our research is as shocking as this emoji 😱, there are only 2 to 5 statistics which are published consistently on a monthly basis that could be considered “real” economic measures. Note from our previous statement that we mentioned “published consistently”, that is, we have dozens of economic variables which are based on a monthly or trimestral frequency, however, very few are published in a timely manner (the table below provides an assessment of publication dates and periodicity of dozens of public datasets). Most of them lag years! The few which are up to date are employment statistics which are published by a U.S. Federal entity -Bureau of Labor Statistics-, Purchasing Managers Index (PMI), and other fragmented sources. Let’s get something straight: this is no novel finding, but is frankly crude and astonishing. The above statement has been published by policy experts in Puerto Rico for more than 20 years and some of them have made this matter their life’s project, so respect to those!

2.0.1 Calendar of Data Publications


Given the above reality, the solution is not limited to criticizing the lack of data, but actually looking for more data. To that end, ABEXUS Analytics, visualized mobility trends, use of internet sources and other non-traditional data sources, to truly assess the current state of the economy. This is a bold statement, since we are aware that measuring the economy is a much more complicated task and some of the alternative data sources will have income biases and limited accuracy for some policy decisions given other structural constraints such as the size of the informal economy. But before we dive deeper into our estimates, let’s highlight how problematic is to have limited performance measures on the economy of Puerto Rico.

3 – US GDP vs PR GNP

3.0.1 US GDP Growth in PR Fiscal Years


The graph above shows the performance of the U.S. GDP over a 60-year period. As noted, if the U.S. National Accounts were to be published on a yearly basis, the U.S. would only have had 2 recessions along this timespan. Without the need to be an economic historian, this is somewhat off! Some of us remember a greater number of not so good old days! However, if we transform that same graph into a quarterly basis, the number of recessionary periods suddenly spikes. In fact, in the case of the U.S. one can note that growth periods are substantially bigger and longer than recessionary ones. As noted in the GNP graphs for Puerto Rico, after 2006, the recession would had been a bigger inverted “mountain” had it not been for the ARRA Funds.

3.0.2 Quarterly US GDP Growth


3.1 PR GNP in Fiscal Years

Puerto Rico has another aggravating issue, which is that GNP data is published on a yearly basis. Imagine how many recessions we probably had during a particular “cuatrenio” that were bluntly ignored! Consider all the policy tools that could have been enacted to curtail such potentially recessionary periods? The issue is simply sad.

3.1.1 PR GNP


Remember the 20-year-old gas gauge analogy at the beginning of this note? We use this adjective (“sad”) because we understand this is not a problem which can be pointed to a single individual or group of individuals. Is a structural problem that is deeply embedded in our institutions and how constituents perceive the value of data. The fact that our main planning institutions have been slashed/understaffed/ or even make fun off, certainly does not help! We know this is flour from another sack… (no worries, we know that idiomatic figures do not translate properly, but the moment required the expression.)

4 – No More Talking, or Tedious Alpha Characters, Let the Data do the Talking

4.1 Job Claims

Job claims data is no rocket science, numbers are astronomically high. The insight behind this is an economic concept known as Hysteresis. Such concept entails many things, but what is relevant to the case of the local economy is that unemployment benefits could cause distortions in the labor markets that must be recognized and tackled. The distribution of salaried incomes is highly skewed towards the lower end. That is, most of the labor force was earning below what is currently provided under the federal unemployment benefits.

From the “bubbles” graph below one can note that unemployment benefits (as of May, 2020) have become Puerto Rico’s largest economic sector, even larger than manufacturing or retail. This will probably remain relevant until the end of the calendar year.

4.1.1 Employment & Weekly Salary of Top Economic Industries


We all have seen and read the news on the huge surge of unemployment claims and all the saga associated with their processing, but if you haven’t had the chance to visualize such a trend in Puerto Rico, the following graph provides a shocking comparison of how big was the jump from March to April 2020.

4.1.2 Unemployment Claims

Initial Claims.html

Most reports have concentrated on the initial unemployment claims, however, the most important figure used to assess changes in the performance of the economy is the “continued claims”. As noted in the graph, -even when the number of initial claims has experienced a reduction and had a peak during the first week of April-, the number of continued claims remain overwhelmingly high (policymakers must keep a close eye on such figures). Thus, there are no signs of a significant recovery in the labor market yet. The growth has been so steep that we expected profound structural changes in our economy, and even the electoral landscape of the coming months… Talking about a leap of faith!

4.2 Labor Force Distribution

The Island’s economic troubles are compounded when the labor force is taken into consideration. In the past few years Puerto Rico’s Labor Force Participation Rate (LFPR) has average between 39%-41%, meanwhile the U.S. rate has remained above 60% even during the COVID-19 pandemic. Now, the LFPR considers those working and those actively looking for work as participating, but for this analysis let’s just look at those that are working. Per ABEXUS’ estimates, only 33% of the civilian population older than 16 is currently working. This means that at best, only about a third of the Island’s population of working age is currently employed. Remember, this isn’t the LFPR. ABEXUS estimates for LFPR is still over 39%, as many are either working or actively looking for employment. Yet, the Puerto Rican economy essentially runs at a third of its full potential. The remainder is either unemployed and looking for work, studying, retired, on disability, or not looking for employment.

Although this may seem like a negative statistic, it does mean that the Island has a labor surplus, but currently lacks the economic opportunities to materialize its potential. ABEXUS estimates take into account the most recently reported unemployment claims, civilian population older than 16, and total employment as reported by Federal Authorities (US BLS LAUS & the PR Department of Labor & Human Resources Household Survey).

4.2.1 Labor Force Distribution by Labor Status

5 – Google’s Mobility Data

Mobility has become almost mainstream due to the infamous contact tracing systems associated with the COVID-19 crisis. Using Google’s datasets (Google Community Mobility index, 2020), we visualized mobility patterns along the period. Such index measures how many “individuals” visit a set of pre-determined locations such as workplaces, groceries and pharmacies, recreation, transit stations and retail stores). As showed in the graph, mobility patterns coincide with the enactment and announcements of the lockdown by the central government. Patterns reflect a drop in mobility which is directly correlated with economic activity, particularly in an economy such as Puerto Rico’s, which depends on cash transactions for consumption and actual presence in a work environment for lower to mid-skill labor. As noted in the graph, Sundays experienced the biggest drop with almost 95% downward change compared to pre-COVID days. Also Retail and Recreation experienced a drop of over 67%, while Groceries and Pharmacies experienced an average drop of 39% between March 15th and April 30th.

5.0.1 Google Mobility


The 35% reduction on groceries and pharmacies is particularly important, since families have inevitably increased their purchases of foodstuffs. However, given the lockdown, Google’s data evidence that average ticket prices must have increased in order to cope with a higher demand for groceries within a single visit to the supermarket. On a previous publication ABEXUS, together with BUREA®, researched and analyzed recent transactional data. Such research validated the above statement, in fact the numbers point towards a 28% surge in the average ticket per visit. Again, this is an example of non-traditional data which provides insights into the happenings of the economy.

Furthermore, a simple look at mobility data by municipality can show which municipalities had a greater compliance with lockdown measures. But there is no need to add more noise to this article 😉. The following table shows which municipalities had the highest drop in mobility associated with grocery stores and pharmacies.

5.0.2 Google Mobility - Municipalities with the Largest Reduction in Workplace Mobility


The following map positions the peak of the lockdown in terms of people not visiting their workplaces during mid-April. After such date, there is an overall tendency of more visits to workplaces. The map also shows that before the enactment of the first lockdown order, several workers were reducing their visits to their work environments. Interestingly, during the first two weeks of March, visits to workplaces tended to increase during the weekends. After such period, workplace visits followed the strict lockdown measures during the weekends

5.0.3 Mobility in the Workplace by Municipality

Red colors indicate less mobility associated with workplaces, while blue colors indicate the opposite- higher mobility

Just to keep your curiosity on the edge, imagine having this data at the time when Puerto Rico spent decades in the legislature debating the famous “Ley de Cierre”!

7 – Correlation Analysis

One of the purposes of this report, was to assess the response of economic actors to the stay-at-home orders, particularly its degree of compliance. Moreover, we also wanted to gauge how people reacted to the number of COVID-19 positive cases that were published on a daily basis.

Well, the short answer is… Yes, people responded with noticeable changes in their mobility patterns!

A correlation analysis carried out by ABEXUS found that there exists a negative correlation between the number of cases reported and workplace mobility. Essentially, given a higher number of positive cases in any given day, the larger the reduction in the number of persons going to their place of work. The correlation is stronger as the number of cases is lagged, all the way up to a week. That is, a week after the number of COVID-19 cases were reported, the correlation with a reduction in workplace mobility was even greater. Given the public perception about the reliability of government data, the above finding is no surprise, people waited almost a week before responding to government data. The following color-coded table presents the results. The green color represents a correlation greater than 80% (very high) and red correlations under 80%.

Somewhat surprisingly, there was no correlation between COVID-19 cases and mobility at Groceries & Pharmacies, meaning that changes in mobility responded to other factors, not just COVID-19 cases. Retail & Recreation mobility did respond to changes in lagged COVID-19 cases. As such, residents were less willing to risk exposing themselves to infection for anything other than foods and/or medicines.

7.0.1 COVID-19 Cases & Google Mobility Correlation


[Technical note] A Pearson Coefficient was used to determine the correlation between Google Mobility indexes and COVID cases and their lagged counter parts (data was lagged between 1 & 14 days). COVID-19 cases were analyzed in a cumulative manner not simply new cases per day. The period of analysis runs from March 15th, 2020 until June 14th, 2020.

8 – The Bottom Line…

Traditional economic measures must not negate the value provided by transactional data from the economy. And again, such analytical skills are not reserved for economists, financiers, data scientists, or even companies like us. Analytics tools have allowed that a myriad of companies, individuals and other institutions appreciate the value of data, since data without context is useless or even worst, data that is not well visualized or communicated, is like writing all this note in Latin and reserving its understanding to a group of preppy boys in a boarding school!

Puerto Rico cannot afford running the economy with a blindfold since the margins of error are somewhat constrained by the fiscal crisis. The efforts should not be limited to pointing out the lack of data, one must understand that technological changes have provided private actors with loads of information which is outside the public realm and must be incorporated into policymaking, or even into business intelligence tools. At ABEXUS Analytics we have recognized such trends and accepted the challenge.

If you or your company is interested in adequately presenting data, if you have an Excel spreadsheet but have not been able to decipher that pivot table or that v-lookup; if you keep using the same lame graphs for you management report or you know a few commands for R-Language or Pyhton, or you simply need some powerful business intelligence, visit www.abexuspr.com We can work something out, as we always take the road less traveled by, which usually makes the difference!

Sources: Employment data: US BLS; Unemployment data: https://oui.doleta.gov/unemploy/claims.asp, Economic Data for PR: the Puerto Rico Planning Board, Statistical Appendix; US Economic Data: FRED, Federal Reserve of the US; Mobility data: Google LLC “Google COVID-19 Community Mobility Reports”. https://www.google.com/covid19/mobility/ . Trends: Google Trends.