Methodology: New York City Economic Recovery Index

Overview

The aim of the New York City Recovery Index is to create a practical, weekly pulse-check of the economic progress of the city since the onset of the COVID-19 pandemic. We built the overall city recovery index from components that track the lives of everyday New Yorkers:

  • Health
  • Transport and mobility
  • Jobs
  • Real estate and housing
  • Restaurant reservations

As New York recovers from COVID-19, the shutdowns, and their economic repercussions, we are tracking each of these elements as they approach, return to, or surpass pre-pandemic levels. To measure the pace of the recovery, we report indices for each component and sum them into an overall recovery index.

We required the following criteria for a dataset to be included in the overall index:

  • Availability: The data must be reported reliably, at least weekly, and available to the public. There are multiple measures that we would have liked to include, but are only reported monthly or quarterly, and often with a significant delay.
  • Relevance: For each individual measure, a return to the baseline (accounting for seasonality) should reasonably reflect the normal functioning of the city. For example, a consistently crowded rush-hour train would suggest that both commuters and tourists have returned to their previous habits.

Data Collection & Analysis

New York City Recovery Index 

The New York City Recovery Index is calculated from an un-weighted average of each of the six sub-indices. A reading of 100 is considered "normal”. The overall index, as well as each sub-index, is constructed so that an increase represents a positive outcome, and a decrease represents a negative outcome. Updates to the index are made weekly, reporting the previous week’s score, and are published by NY1 as well as Investopedia.

COVID-19 Hospitalizations Index

The COVID-19 Hospitalizations Index represents the impact of the pandemic on the functioning of the city.

Data

The daily count of hospitalizations with an associated positive COVID-19 test are taken from New York City’s Department of Health and Mental Hygiene. We choose to base our health index on hospitalization counts rather than positive COVID-19 tests or deaths. Testing data is dependent on the number of people who can be tested, which fluctuates based on testing supplies available, distribution of those supplies to medical professionals, and testing facilities open to the public, as well as individuals’ decisions to be tested.  Death data often considerably lags the spread of the virus.  While the reported number of daily hospitalizations with an associated positive COVID-19 test may overstate the number of hospitalizations caused by COVID-19 infection, we concluded that they are nonetheless the most stable measure to track the effect of the virus on the city.

Source

COVID-19 Daily Counts of Cases, Hospitalizations, and Deaths Provided by the Department of Health and Mental Hygiene (DOHMH)

Transformations

The aim of this index is to understand where we are in the progress of containing the impact of COVID-19, as it relates to the functioning of New York City. Therefore, we built the index to reflect the following: When there are no daily hospitalizations, the index will be 100. In April, when New York City passed 1,500 hospitalizations per day and city officials had enacted the strictest business closures to date, the index should be close to its minimum possible value (0). Therefore, like the initial unemployment claims index, as the number of hospitalizations goes down, the index will increase back toward 100.

One of the main concerns impacting the measures to contain COVID-19 is the rate at which it spreads. Therefore we are concerned with the rate of increase (e.g., whether or not hospitalizations double week-over-week) as much as the total number. To construct the index, we took the trailing 7-day average of daily hospitalizations beginning on February 29, 2020. We based the index on the log of the number of hospitalizations to ensure the index directly tracks the rate of change of hospitalizations.  

This aligns with the measures that health officials advise for social distancing practices and thus the viability of much economic activity in New York City. Because there were no recorded COVID-19-associated hospitalizations to begin 2020, we set the baseline for early weeks to ‘100’. For the component score, the COVID-19 Hospitalizations Index is divided by six to be equally weighted alongside the other components.

Subway Mobility Index

The New York City subway is a powerful indicator of movement around the city, and the physical reality of this mode of transportation also means ridership has been especially impacted by the pandemic. 

Data

The index is composed of MTA data showing day-by-day ridership numbers for the New York City subway. The MTA calculates the percent change against the ridership/traffic volume from the equivalent day in 2019 (generally the same day of the week during the same week), with data beginning on March 1, 2020. For index values occurring before March 1, we used MTA data showing the number of turnstile entries and exits made each week by customers in each station of the New York City Subway. The counts are aggregated from readings occurring every 4 hours from turnstiles in subway stations across the city. The MTA publishes these observations on a weekly basis, covering seven-day periods beginning on the Saturday two weeks prior to the posting date and ending on the following Friday.

Source

MTA Day-by-Day Ridership Numbers

Transformations

To construct the index, we use the 7-day trailing average of the daily percent change from the equivalent pre-pandemic (2019) day in subway ridership as now reported by the MTA. These values are converted to whole numbers for the Subway Mobility Index. The MTA reports these daily ridership numbers going back to March 1, 2020.  

To construct the index prior to March 1, 2020, we calculated the same percent change from the weekly turnstile reports. Observations were compiled starting in December of 2018 in order to establish a historical baseline. We calculated the total daily entries by station and turnstile and then summed across all the stations to get daily subway entries for all of New York City since late December 2018. For the component score, the Subway Mobility Index is divided by six to be equally weighted alongside the other components.

Home Sales Index

We believe home sales to be a measure of economic confidence in the city on behalf of homeowners and real estate investors. Furthermore, home sales often involve large financial products which bear significance beyond the individuals transacting the sale. As a measure, the Home Sales Index represents the volume of pending home sales in a given week in New York City. 

Data

The index is created from StreetEasy data showing weekly pending home sales. The data are collected from StreetEasy’s online home listings for the city of New York. A sale is marked as pending by the listing agent when the seller has accepted an offer from a buyer; however, it does not indicate a closed deal. Deals may not close for a variety of reasons including inspection, financing, etc. Because an agent need not necessarily report a closed deal, or may choose to do so long after the deal has closed, though, we have chosen to use data on pending sales (which are recorded on the StreetEasy platform in a timely and reliable way) as the most reliable measure.

Source

StreetEasy

Transformations

We calculate percent change relative to a pre-pandemic (2019) baseline in pending home sales in NYC from weekly reports provided by StreetEasy. To create a pre-pandemic baseline, we average the equivalent week’s sales volume with the week before and the week after. We use this three-week rolling average to mitigate the effects of sharp weekly changes while maintaining a seasonally-appropriate point of comparison. Year-over-year percentages are converted to whole numbers for the Home Sales Index. For the component score, the Home Sales Index is divided by six to be equally weighted alongside the other components.

Rental Inventory Index

While pending home sales are an indicator of the economic confidence of buyers, rental inventory reflects the lived conditions of the majority of New Yorkers for whom buying a home is cost-prohibitive. We use rental inventory because the number of available units reflects the changing conditions of the market before landlords adjust the asking price. 

Data

As a measure, the Rental Inventory Index is derived from the number of homes available for rent in a given week in comparison to the baseline number of homes available to rent in an average month, seasonally adjusted. The data are provided by StreetEasy and collected from online rental listings for the city of New York. 

Source

StreetEasy

Transformations

There are two extremes when the rental market indicates a perilous situation for the NYC economy. If there are too many apartments listed for rent, this may indicate a lack of demand, suggesting people may be leaving the city, or that there is a disconnect between available inventory and the type of inventory needed by New Yorkers. If there are too few apartments available for rent, prices may rise and renters may be priced out of the rental market. To account for both scenarios, the rental index measures how far the rental inventory is, in any given week, from a historically-averaged range.

For this index we track weekly rental inventory figures against a predictive seasonal model created from 2010-2019 monthly data. To create the model, we used rental inventory numbers from the past 10 years to determine typical monthly variation from January, including each year’s overall growth or decline.  

We then compare the model to the present by measuring the growth or decline of each week relative to the given year’s January average. The difference between the predictive model and the current data allows us to see how unusual each week is compared to the past 10 years. The index is then created to reflect the difference between the model and the current weekly figures, and higher index values are intended to reflect market conditions closer to historical averages. So, if there are too few rentals, the index goes down, and if there are too many rentals, the index also goes down. For the component score, the Rental Inventory Index is divided by six to be equally weighted alongside the other components.

Unemployment Claims Index

The Unemployment Claims Index represents the employment health of New York City. An unemployment claim is an application for cash benefits that an employee submits after being laid off or being unable to work for other covered reasons, such as the COVID-19 pandemic.  As a measure, we use initial claims for unemployment insurance (UI) to track the economic recovery of the city. A rise in initial unemployment claims, seen as a decrease in the index, began the week of March 21, 2020—a little over one week after a national emergency was declared for COVID-19.

Data

In March 2020, the New York State Department of Labor (NYSDOL) Research & Statistics Division began publishing weekly data on initial unemployment insurance claims in New York City, including the net change from a pre-pandemic (2019) baseline. For the first 12 months, these data were published as reports by the Research & Statistics Division. After a pause in reporting, weekly initial claims data are now available again by selected characteristics as part of their Occupational & Industry data product.  

To account for this pause in reporting, articles from March 2021 until January 2022 reference changes to the index derived from estimates for city-level initial unemployment insurance claims that are imputed from the weekly statewide reports provided by the U.S. Department of Labor. 

Sources

New York State Department of Labor, Weekly Unemployment Insurance Claims Reports
United States Department of Labor Unemployment Insurance Weekly Claims Data

Transformations 

The index takes the inverse of the percentage change from the pre-pandemic baseline (2019) of unemployment claims. For example, if unemployment claims this year are 2X higher than last year (200% year-over-year), then the index is 50 (100/2). If initial claims are 4X higher than last year (400% year-over-year), the index is 25 (100/4). If initial claims are the same as last year, the index will be 100 (100/1). So as initial claim numbers trend back to the level seen in 2019, the index will increase back up to 100.

From March 2021 to January 2022, the weekly initial UI volume was imputed from the statewide data to account for a pause in reporting the New York City weekly data from the NYSDOL. During this time, we applied the estimated proportion of the city’s initial UI claims to the weekly reports of total initial UI claims for New York State. In order to calculate the proportion of the city-to-state claims, we calculated a three-week rolling average (including the previous, equivalent, and following weeks) to the weekly city and state data from 2020. This allowed us to preserve the seasonality of initial UI claims while smoothing projection and mitigating sharp weekly changes. Models using older historical data were considered but ultimately discarded in favor of a model based on 2020 as the impact of the pandemic on the city relative to the state was uncharacteristic of the recent past. For the component score, the Unemployment Claims Index is divided by six to be equally weighted alongside the other components.

Restaurant Reservations Index

Restaurants are a key indicator of the economic life of New York City, and are especially sensitive to the fluctuations of the COVID-19 pandemic. 

Data

The index is created from OpenTable data showing daily seated diners compared to a pre-pandemic baseline. While not all restaurants take reservations, OpenTable hosts reservations for over 39,000 restaurants in the New York City area.  The data are collected from a sample of restaurants on the OpenTable network and include online reservations, phone reservations, and walk-ins. Day-of-week fluctuations are accounted for by comparing the same day of the week from the same week pre-pandemic. OpenTable data reporting begins on February 18, 2020. A baseline index of 100 is assumed for January 1st-February 17th, 2020. 

Source

OpenTable

Transformations

Year-over-year percentages are converted to whole numbers for the Restaurant Reservations Index. For the component score, the Restaurant Reservations Index is divided by six to be equally weighted alongside the other components.

Retired Sub-indices

New Businesses License Index

From July 27, 2020 to September 7, 2020 we included a New Businesses License Index built from data provided by the Department of Consumer and Worker Protection (DCWP) via NYC OpenData. The DCWP licenses more than 75,000 businesses in more than 50 industries (overview of businesses licensed by DCWP) and updates the dataset with new business licenses weekly. We decided to discontinue use of New Business License data as a part of our index because the small base sizes introduced too much variability to track weekly changes, and we found the department’s coverage of small businesses too limited to serve as a functional measure of the health of small businesses in the city.

Current Limitations and Areas for Future Exploration

There are countless ways of measuring the economic impact of COVID-19. The New York City Recovery Index is not meant to be a complete portrait of every economic aspect of an economy as diverse and multi-layered as that of New York City. There were several measures not included that fit our objectives, but are not consistently or publicly available.The NYC Recovery Index does not reflect the economic recovery of any individual. The iniquitous distribution of wealth, as well as the disproportionate impact of the pandemic on already marginalized residents, results in a broad range of lived experiences not reflected in the index. We hope to be able to supplement our weekly reporting with richer data on a monthly basis.

We further acknowledge that with each data set there are limitations. Initial unemployment insurance claims, for example, do not reflect the job losses of undocumented individuals who represent an important and especially vulnerable segment of the workforce in New York City. Additionally, the accuracy of initial COVID-19 hospitalization data was impacted by the availability of tests in the early days of the pandemic, among other factors, and the accuracy of more recent COVID-19 hospitalization data is impacted by the significant number of hospitalizations resulting from non-COVID-19 causes but which associated with a positive COVID-19 test. We also use data provided by companies such as StreetEasy and OpenTable, which service a sizable portion of the real estate and restaurant industries but which cannot account for all observable pending home sales, homes for rent, or restaurant reservations in New York City.

This index is not meant to serve as a predictive measure of how long it will take for the city to recover; rather, it is a weekly readout of where the city is compared to the pre-pandemic economy. One limitation of measuring recovery in this way is that it is possible that the future economic landscape of New York City will differ from its past, and require new indicators to better reflect that reality. For example, as new strains of the COVID-19 virus continue to emerge among an increasingly vaccinated population, public health officials may choose to loosen recommended restrictions and protocols to contain the spread (as with the CDC’s shortened recommendation for isolation after a confirmed COVID-19 case).  For this reason among others, we regularly revisit and update our methodology and approach to the index.

Acknowledgements

This index was created by the Investopedia Research Team with direction from Jon Roberts, Ph.D., and in collaboration with Alexandra Kerr, Caleb Silver, and Dylan Zurawell.

Special thanks to Spectrum News NY1 for their collaboration and partnership. Special thanks also to StreetEasy for providing weekly pending home sales, historically and by borough, and for their collaboration and clarification in the use of their data.