Sizing markets.
Sizing Pakistan’s e-commerce foot print as at year end 2022 (31-Dec-22) using a mix of methods and data sets.
Instructions
Estimate current and projected e-commerce market size for Pakistan as at 31–12–2022. Project market size for 31–12–2023. Identify highlights and trends for internal strategy discussions.
Data Sources:
SBP has been publishing payment systems reports since 2005. Use dataset from published SBP payment systems reports from 2017–2022.
Objective:
Collate data from SBP payment systems report in one table. Use SBP table to estimate current market size, review growth assumptions and project expected market size 12 months down the road as at 31–12–2023.
Design:
Use payment system data to estimate payments made via cards on ecommerce orders. Plug in assumption for proportion of digital transactions (payment made via a digital payment option) versus cash on delivery (CoD) transactions. Estimate current and projected market size for ecommerce transactions in Pakistan in USD. Clearly state and validate all assumptions.
Solved solution
Tl;dr;
Current numbers. Source. SBP payment systems report.
Projected numbers. Extrapolated trendline using historical figures.
31.12.22. Digital payments for ecommerce. USD 609 million for the year.
31.12.23. Projected estimate. USD 700 million.
31.12.22. Ecommerce market size estimate. Pakistan. USD 4 to 6.1 billion.
31.12.24 Ecommerce market size estimate. USD 4.6 to 7 billion.
Market summary. 15 million mobile banking application users. 6,000 registered ecommerce vendors. 36 million transactions per year. Avg. transaction volume per user per year. USD 67.
Estimation Models.
We have two primary models types that we can use to answer this question.
a) Direct data driven models. b) Relative comparison models.
Data driven models work with direct market data. Relative comparison model ask a different question. If Pakistan’s GDP is X and Indonesia GDP is Y and Pakistan is 3 years behind the curve compared to the Indonesian market, can we estimate Pakistan’s foot print, given Indonesian figures?
Alternatively if Pakistan’s GDP is between $376 to $400 billion, retail is generally 18%–20% of GDP and ecommerce is between 5% — 15% in emerging and frontier economies, what would be Pakistan’s ecommerce footprint, given Pakistan’s GDP?
For direct models, direct market data would be recorded digital payments through the banking systems, captured and reported via the SBP payment systems report.
But digital payments represent one part of the market. The other part is Cash on Delivery (CoD). While there is no direct recoded disclosure of CoD trends, we can speak to acquirer banks, digital vendors and logistics partner to get a sense of their estimates.
We can also use hybrid approaches where we can validate numbers from one approach by using data from an alternate approach.
Solution design.
- Collate SBP payments systems data set.
- Estimate total digital payments received for ecommerce orders.
- Plug in estimate for share of Cash on Delivery (CoD).
- Estimate total ecommerce market share for Pakistan.
Data.
The data set was created by capturing 13 fields from SBP quarterly payment systems reports. In addition to SBP payment systems report, a history of USD:PKR exchange rate was used to calculate the average quarterly exchange rate for each reporting quarter.
USD equivalent values and historical exchange rates.
Figures in PKR were converted into USD equivalent using applicable exchange rates. We used published SBP USD:PKR day end re-valuation exchange rate. The daily exchange rates were used to calculate a quarterly average for the time period. PKR payments were converted into USD equivalent amount using the quarterly average for each quarter.
Quarterly conversion averages out the impact of PKR depreciation against USD and is more accurate than annual conversion using a single rate for the year. A single rate applies a year end rate to all transactions throughout the year. A quarterly average rate applies a more appropriate rate to convert transactions to USD.
Digital payments.
We assume ecommerce payments refer to payments made via cards (credit, debit, pre-paid) using an internet payment gateway (IPG). Payments made via other digital products using the same IPG are also classified as digital. Payments made outside of IPG (using IBFT, CoD, RAAST, or mobile money transfers using cash deposit) are classified as CoD or non-digital. All pre-payments made outside of IPG are also treated as CoD.
Real growth.
All estimates and measures of growth are calculated in USD terms. While growth estimates in PKR terms are also relevant, USD growth rates are a reasonable proxy for real growth and reduce the impact of inflation and nominal growth. They also make comparison with other economies easier.
Registered stores.
Stores with an IPG registered with a local bank or payment solution provider as an ecommerce store.
Baseline digital ecommerce payments.
Establish baseline for ecommerce payments. Use baseline to establish current transaction value as well as projected future growth rate.
Annual estimates use SBP quarterly estimates for the calendar year in question. 5 year trend is used to establish baseline.
For the year ended 31–12–2022, digital ecommerce payments in Pakistan crossed USD 609 million for the full year.
Our dataset starts with Q4 2017, so we can’t use 2017 figures for comparison because we only have one out of the four quarters.
Only compare ’18 to ’22 calendar year when evaluating annual growth.
While Q4 for every year tends to be the biggest quarter of the year, this was no longer true in 2022.
Quarter on quarter growth for ’22 was mostly flat. Growth was also flat when comparing Q4 ’21 with Q4 ’22 results.
2022 is the first time Q4 yoy growth was flat in 6 years.
Digital ecommerce grew from USD 181 million in ’18 to 267 million in ’20. Then more than doubled to USD 609 million by year end ’22.
While overall market size has grown by 100% in two years, growth in ecommerce orders and digital payments has been uneven.
The years with big jump were ’20 and ’21. Covid-19 did more to change purchase behavior of domestic consumers then break through technology or marketing ever could. But then global consolidation wave at start of ’22 led to a funding crunch and economic slow down in Q3 and Q4 ’22 put the brakes on growth in Pakistan.
8 quarters of positive growth followed by a mostly flat year that still managed to grow at a 20% clip in dollar terms. Growth is more likely to be flat or slower than historical averages in ‘23.
Beyond growth there are two key trends to watch and track.
Average transaction value per user in USD. The drop in average transaction value per user from 2017 to 2022.
The annual average for total ecommerce transaction value per user has fallen from USD 188 in ‘17 to USD 67 in ’22.
This was bound to happen as ecommerce goes mainstream. The effect is more visible because of significant depreciation of the Pakistan rupee against the US dollar. But it would have still happened over time even without the depreciation as customers and vendors moved from buying big ticket electronic items to lower value weekly grocery baskets.
Reduction in average gross merchandise value (GMV) processed by e-commerce stores. The second effect is a result of growth in ecommerce stores. As new stores get added in large numbers, they start with a smaller base. Overall average sales per store fall with the new additions. Every new store is not J., Khaddi, Ego, Sana Safinaz or Gul Ahmed Ideas.
As new stores get added average GMV is likely to fall lower. Projected outlook for 31–12–2023 year end store count using historical trend is between 8,000 to 8,500 registered stores.
The numbers are likely to grow faster in ’23 because of the push by PSP, EMI and digital payment providers as well as instant e-commerce store solution providers. The two groups together have done more for adding ecommerce vendors to the local eco system than any other driver.
Projected store count and an adjusted lower GMV per store can also be used to predict and forecast year end ’23 ecommerce digital payments.
Projected ecommerce digital payments for year end 31–12–2023 using historical trend are likely to be $650 to $700 million for the full year.
Share of Digital payment vs CoD.
If we assume digital payments represent a small fraction of actual ecommerce payments across Pakistan, we can use a simple model to estimate the total market size.
Divide digital payments by the share they represent.
For instance if digital payments are USD 600 million and the share is 10%, 600 / 10, will give us a USD 6 billion market size. This is the figure (USD 6.4 billion) our friends at Data Darbar recently reported in their sector brief.
Here digital payments refer to Card Not Present (CNP) transactions where a credit or debit card is used to process a payment without the physical card being present.
Estimating ecommerce market size with share of CNP transactions.
Depending on who you speak with, the portion of digital and prepaid transactions is estimated between 15% — 30%. The estimate varies if you speak with banks, vendors, logistics providers or payment solution companies. The plot below shows the range of likely estimates for combination of assumptions.
The vertical axis in the plot above (range 400–900) represents the estimated digital payments received for ecommerce transactions for the full year ’23.
The horizontal axis represents share of CoD (range 90% — 60%) in ecommerce payments in Pakistan.
The corresponding figure at the intersection of these two axis represents the size of the ecommerce market for any given combination of these two values.
Historical outlook
How has the outlook changed over the last 5 years?
In March 2020 as part of a market sizing exercise for a client we had predicted year end ’22 digital ecommerce payments at PKR 133 billion (USD 782 million at the then USD:PKR exchange rate).
While we are on the money in PKR terms we are off by USD 180 million in USD terms.
In Dec ’21 / Jan ’22 as part of three separate engagements we had estimated a broad range between USD 2 billion to USD 8 billion for the total ecommerce market in Pakistan.
The current range at year end ’22 is roughly in the same ball park.
If we annualize SBP quarterly figures and assume a steady digital component of 15% (85% CoD) here is what sizing ranges for the local ecommerce market have looked like across 5 years. As we change the CoD% the numbers will change.
Market player estimates range from 75% — 85% CoD share of ecommerce transactions. CoD share is likely to drop further as confidence in IBFT transfers as a pre-settlement mechanism improves. However all IBFT transfers may not reflect in digital ecommerce payments depending on how these transactions are tagged or tracked.
Supportive trends. Growth in mobile banking users.
Two plots to showcase growth of mobile banking application users.
One. 15 million users as of year end ’22. Projected 16 million users year end ’23. Growth continuing at a steady clip. It showcases appeal of mobile banking use case, availability of banking applications and promise of a ready market for digital banks as well as digital payment solutions. Deployment of retail banking middle ware in banking tech stacks, availability of cheaper smart phones and 4G networks have all helped this trend grow.
Two. The steady rise in the average transaction volume per user per quarter in USD terms. Almost 5x growth in the figure over 5 years. While internet banking usage is distributed across corporate and retail accounts, mobile banking usage is assumed to be mostly retail customers.
Once again showcases customer comfort with digital payment use case as well as broad acceptability of this channel as a payment medium.
I teach the Tech product development and Applied Data Analysis courses at IBA Karachi. This post is based on the notes from an in class market sizing exercise.
If you enjoyed reading this post, you may also like
