Every day, the Instawork platform handles thousands of transactions involving businesses and hourly professionals, generating a huge amount of data on hourly pay as well as other aspects of the labor market. This report summarizes some of the major trends in demographics, roles, and worker constraints in regions across the United States.
To receive future economic insights from the Instawork Economic Research Division, please subscribe by visiting: https://hubs.li/Q012kZ--0
The data are also compiled into the Instawork Pay Signal Index (PSI) and indexed trends in hourly pay. Because businesses can book shifts in advance on the Instawork platform, the metrics include forward-looking data for the current month as well. Please refer to the appendix for explanations of the methods behind each metric.
A decent majority of businesses raised pay in July, after advance bookings showed pay trending downward. Businesses are again trying to lower hourly pay in August, but the same sequence of events could repeat:
Our in-app surveys track Pros' labor market situations on a monthly basis. For details on methods and questionnaires, please see the Methodological Appendix below.
Here is the distribution of Pros by the jobs they worked up through the reference week (17 July 2023) this month:
Work in full-time and part-time jobs rose in July, as Pros showed a slight shift in preference toward permanent positions rather than app-based or temporary work:
This is a small change, but the greater numbers of Pros seeking and attaining full-time and part-time jobs could suggest precautionary behavior as a result of uncertainty about the labor market. More Pros also wanted no job, with the share off Pros working 0-10 hours increasing again in July. In fact, all other categories besides 31-40 hours shrank slightly during the month:
This was in opposition to the hours Pros wanted to work – a smaller share wanted 0-10 hours than in June, even though many more actually did:
Summer vacations could interfere with these numbers somewhat; we will have a better idea of the health of the labor market in September.
In July we also asked Pros about their health insurance, including private insurance coverage via relatives:
More than 70% of Pros say they are covered by health insurance of some kind. Eligible Pros who work shifts with W-2 tax treatment are also offered insurance. Still, the most common type of insurance among our Pros is provided by the government. A smaller share of Pros said they had purchased insurance themselves than the national average of 10%. As flexible work becomes more common, the option to purchase affordable insurance is likely to become more important.
Because flexible work is one of several options that workers might have in the labor market, increases in flexible work may mean decreases in other areas. The following statistics measure differences in shift work booked on the Instawork platform (measured in hours), month over month:
Demand for food service workers continued to climb steeply in July, as did demand for other leisure and hospitality roles. The exception was private events, where demand was softer again. The glut of private events in 2022 may have crowded out the demand in 2023 somewhat, while large-scale entertainment in stadiums and arenas keeps growing.
Technical note: To control for the growth of the Instawork platform, only business locations that have participated for at least two months before the start of the comparison period are included. Changes in hours are included only for roles for which businesses booked shifts during both months.
The share of men working shifts on the platform increased for the first time since February, and the share of Black/African-American workers dropped for the first time since March:
A sustained increase in men using the platform might signify a worsening of their other options in the labor market. But right now it is too early to say whether these trends signify a shift in the broader economy.
Except in building and grounds cleaning, a slight majority of businesses raised pay in July. In the sectors with the biggest increases, businesses are trying to push pay rates back down in August. These reductions are most likely to stick in hospitality, where the summer season is winding up in many markets:
Hourly pay for concession or stand workers and dishwashers rose slightly in July but was steady for other hospitality roles. In August, concession and stand workers may return to their peak pay from May 2022, when a surge of spending on leisure and hospitality sent rates steeply higher. Pay for entry-level warehouse roles also ticked up, but pay for intermediate warehouse and general labor roles stabilized:
To receive future report briefings or data insights from our Economic Research team, please subscribe below.
Five surveys per month are delivered via the Instawork Pro app. Random samples of Pros who have worked at least one shift in the past 12 months may be shown a one-question survey when they open the details for a shift. No Pro receives more than one survey per month. The surveys continue until they collect 1,000 responses.
The repeating questions on Pros' labor market situations are as follows:
1. Please mark all the kinds of work you will do this week:
2. Please mark all the kinds of work you would like to do each week:
3. In total, how many hours will you work this week at all your jobs?
4. In total, how many hours would you like to work each week at all your jobs?
The PSI gauges the overall direction of changes in hourly pay from month to month, much the way a purchasing managers’ index measures supply chain activity.
To begin, for each month, we measure the average hourly pay offered by each business on our platform for each role in each region. If the business offered shifts for the same role in the same region during the previous month, we record whether the average rose or fell. A rise is recorded as +1, a fall as -1, and no change as 0.
Next we weight this signal by the average number of shifts the business offered for that role across the two months. For example, if the business offered 10 shifts for line cooks in the Houston area during February and 18 shifts for the same role in the same region during March, then the weight would be 14.
We group these weighted signals by the Census Bureau’s occupational categories and take the weighted average for each category. Then we multiply the weighted average by 50 and add it to 50. This yields a PSI between 0 and 100. At 0, the PSI implies that all businesses in the sample offered lower pay. At a PSI of 100, all businesses offered higher pay. At a PSI of 50, businesses that raised or lowered pay did so with equal weight (or loosely, equal numbers of shifts).
At present we cover six major occupational categories. These are the numbers of workers they represented in the Bureau of Labor Statistics’ counts of hourly workers in 2022, and their shares of the total reported number of hourly workers:
To create a national PSI, we calculate a weighted average by weighting the PSI for each occupational category by its share of hourly workers above. The resulting national PSI represents occupational categories that cover roughly 44% of hourly workers in the American labor force.
Sample selection for our pay trends is similar to the method for the PSI. Businesses that book shifts for the same role in the same region during consecutive months are the units of observation. For each pair of months, we calculate the change in the average hourly pay offered for the given role in the given region. Examples of roles are line cooks, forklift drivers, and custodial staff.
Next, as for the PSI, we weight the changes in pay by the average number of shifts across the two months. Then we calculate a weighted average of the changes at a national level for each role. To create an indexed trend, we have chosen July 2021 as the starting point, where the indexed hourly pay for each role is set to 100. We then use the monthly changes to map the trend from August 2021 onward.
At present we publish the indexed trends in hourly pay for 14 of the roles staffed on our platform:
