Mexico’s remittances and the household purchasing-power connection

An international remittance is measured in several economically meaningful ways. The sender gives up income or savings. A payment provider moves funds across borders. The recipient receives pesos or another settlement asset. The household then decides how much to spend, save, invest or use to repay debt. A dollar increase at the border does not guarantee an equal increase in the recipient’s real purchasing power, and a decline in transfers need not reduce current consumption one for one if the household has other resources.

Mexico’s remittance statistics make this distinction consequential. Banco de México’s February 2026 release reported receipts of $61.791 billion for 2025, compared with $64.746 billion for 2024, a decline of 4.6 percent. These are the historical values in that release, rather than an assertion that no subsequent revision occurred. The same release records an important difference between transmission and collection: 99.1 percent of receipts used electronic transfers, while 49.6 percent of that electronic value was collected in cash. Electronic cross-border delivery does not imply that every beneficiary retains a deposit balance. Banco de México, remittances for December and full-year 2025, pages 1–2.

More recent evidence also requires precise periods. The October 1, 2026 release reports $41.802 billion during January–August 2026, up 2.2 percent from the same months of 2025, while August alone declined 3.6 percent from a year earlier. For that month, fewer transactions and a higher average transfer moved in opposite directions. The cumulative increase and monthly decline can both be correct. Neither establishes how many distinct households gained or lost income. Banco de México, remittances for August 2026, pages 1–3.

The analytical question is how a cross-border transfer becomes durable household resources under changing exchange rates, prices and sender employment. This report develops a transparent accounting bridge and examines the limits of the available aggregate and household evidence. It does not estimate a new national consumption multiplier, forecast the peso or infer the behavior of every recipient from an average transaction.

For a particular transfer, spendable pesos depend on the amount available for conversion, the exchange rate actually applied and any recipient charges. Real purchasing power also depends on the prices of the goods and services the household needs. A sender’s total payment can include a fee paid in addition to the transfer principal, or a fee deducted from a fixed sending budget. Those conventions must be specified before comparing offers or measuring a cost reduction.

Consider an explicitly hypothetical fixed sending budget of $300. If a $5 fee is deducted before conversion and the recipient receives 18 pesos per dollar, the delivery is 5,310 pesos. At 20 pesos per dollar with the same fee convention, delivery is 5,900 pesos. The second amount is 11.1 percent larger in nominal pesos. If the relevant local price level rises by five percent between the two observations, the increase in purchasing power is only about 5.8 percent. The exchange-rate change and inflation cannot be interpreted independently when describing the household outcome.

The actual conversion rate matters. An official reference rate is useful for a consistent aggregate index, but it is not necessarily the rate offered by the provider at the moment of settlement. A transfer may be priced before or after the sender initiates it. Promotions, payment method and delivery channel can change the effective rate and fee. A national calculation using average reference rates is therefore a standardized purchasing-power indicator, rather than a direct observation of every beneficiary’s receipts.

The household account extends beyond remittances. Domestic wages, business income, public transfers and investment income determine total recurring resources. Borrowing and withdrawals from savings can finance spending without representing new current income. A family maintaining consumption by drawing down deposits after a remittance loss has stabilized expenditure while becoming financially less resilient. Consumption alone can consequently understate the cost of the shock.

For welfare analysis, the account should also record household size, health expenses and existing commitments. A transfer divided among six members does not provide the same resources per person as an equal transfer to a single adult. Housing status changes the relevant budget pressures. A common transfer amount is an entry in the resource account, not a sufficient description of living standards.

An aggregate approximation can hold fees and settlement conventions constant and combine three indices: dollar remittance receipts, pesos per dollar and domestic prices. The real peso index equals the dollar-flow index multiplied by the exchange-rate index and divided by the price index. This is an identity for the specified standardized index. It does not establish that the exchange rate or remittances were causally independent.

The following scenario is hypothetical and uses no observed Mexican growth rates. Dollar receipts rise three percent, the peso appreciates so that pesos per dollar fall ten percent, and domestic prices rise five percent.

The exact calculation is 1.03 × 0.90 ÷ 1.05, giving a real decline of approximately 11.7 percent. Simply adding the three percentage changes would produce an approximation and omit interaction terms. For larger movements, the difference becomes economically relevant. Log changes provide an additive decomposition, but an analyst should identify that method rather than presenting log points as ordinary percentage changes.

The direction of the exchange-rate quote must be stated. A decline in pesos per dollar means the same dollar transfer converts into fewer pesos. It does not mean that every Mexican household becomes worse off: cheaper imported goods can benefit consumers and firms, with the extent depending on price adjustment. The relevant net outcome differs between a remittance recipient buying locally supplied services and a business purchasing imported equipment.

Period matching is equally important. A monthly dollar flow should be converted with a corresponding exchange-rate measure and deflated with a price index for the same period. Multiplying an annual flow by the exchange rate on the final day of the year can distort the annual purchasing-power estimate. Where monthly flows are uneven, the sum of monthly conversions can differ from an annual total multiplied by an unweighted annual exchange-rate average.

Finally, the national consumer-price index represents a standardized basket. A recipient’s costs can rise faster or slower because food, rent, transport, health and education have different weights in that household’s spending. A national real-remittance index is informative for broad purchasing-power trends; it is not a household-specific cost-of-living measure or a direct measure of poverty transitions.

Total receipts equal the number of transactions multiplied by the average transaction amount, subject to the source’s units and rounding. This decomposition can identify whether aggregate changes occur through frequency or transfer size. It cannot identify the number of unique senders, recipients or households. One migrant can send several transfers to one household, several family members can support the same recipient, and one person can receive money intended for a wider family group.

A hypothetical example illustrates the identification problem. Initially, 100 households each receive two transfers of $200, producing 200 transactions and $40,000 in total receipts. Later, 80 households each receive two transfers of $250. Total receipts remain $40,000, the average transaction rises and the number of transactions falls. Twenty households lose their transfers entirely. An unchanged total conceals a large distributional change.

Another allocation could produce exactly the same later aggregate: all 100 households receive money, but some consolidate several smaller payments into fewer larger transactions. That change could reduce fixed fees without reducing participation. Transaction data alone cannot distinguish these cases. Statements that a fall in payment counts means an equal number of families stopped receiving support require a recipient-level panel or another defensible identification method.

Survey data introduce a household perspective, with their own coverage and measurement limits. INEGI’s ENIGH 2024 results report average quarterly current household income of 77,864 pesos and average income from other countries of 598 pesos across all households. The latter is not the average transfer among recipients. The table also includes pensions, government benefits and other transfers as separate categories. Treating the entire transfer category as international remittances would materially overstate the relevant income source. INEGI, ENIGH 2024 results, pages 4–6.

A small mean across the full population can coexist with high dependence among recipients. If, hypothetically, four of every 100 households each receive 15,000 pesos in a quarter and all others receive zero, the population mean is 600 pesos. The conditional recipient mean is 15,000. Neither is an error; each answers a different question. Their difference is precisely why a national average should not be presented as a typical beneficiary’s resources.

Distribution also depends on whether the statistic averages household shares or divides aggregate remittances by aggregate household income. A low-income and high-income household receive different weights in those calculations. Reporting the denominator and weighting rule is essential when assessing exposure. A ratio can describe importance in household budgets without establishing whether remittances caused the observed income level or expenditure pattern.

Aggregate cross-border statistics and household surveys are collected for different purposes. The former measure flows across an economic boundary, using reported payment activity and estimates for relevant channels. The latter describe sampled households’ income and expenditure, with questionnaire definitions, recall periods, sampling weights and respondent behavior. Their totals can differ because of underreporting, coverage, timing, concepts and valuation. A discrepancy is a research problem to investigate, not a license to replace one series silently with the other.

Banco de México’s remittance releases explain that direct cash and in-kind estimates draw on INEGI’s international traveler surveys and are revised as final information becomes available. This identifies a specific estimation and revision process. It does not imply that every difference between the balance of payments and ENIGH arises from that channel. The electronic share describes how measured transfers cross the border, while the household survey records reported receipt within its own design.

The survey is especially useful for linking reported foreign income to household composition, other income sources and spending. It is less suitable for establishing the exact national transfer total without a reconciliation exercise. Conversely, the national flow series does not reveal which income decile receives the money or who controls it within a family. Combining the strengths of the sources requires retaining their distinct units and coverage.

A proportional adjustment that scales every surveyed recipient’s remittances up to match a larger macroeconomic total embeds a strong distributional assumption. It assumes the missing amount is allocated in the same proportion as the reported amount. Missing high-value recipients, omitted small transfers and systematically underreported receipts could imply very different allocations. The resulting adjusted poverty or inequality estimate would be conditional on that allocation, not an independently observed household outcome.

A better reconciliation would document concepts first, then examine the distribution of reported amounts, receipt frequency, payment methods and survey nonresponse. Alternative allocation scenarios can show how sensitive conclusions are to missing information. They should remain labeled scenarios. This report does not perform a microdata reconstruction or publish a revised national recipient count.

A visible commission is only part of the cost. The World Bank’s Remittance Prices Worldwide methodology separates transfer fees from the exchange-rate margin and identifies payment and delivery arrangements. It also describes a collection snapshot rather than a continuously executable quote. Its standard price points cover amounts corresponding to $200 and $500, with methodological provisions for local-currency equivalents. These observations are useful for comparing defined services, but they do not establish the actual price paid on every transfer. World Bank, Remittance Prices Worldwide methodology.

For illustration, suppose a provider charges a $3 fee on a $200 principal and applies an exchange-rate margin of two percent relative to a common benchmark. Under a simplified convention in which the fee is additional to the principal and the margin is measured on the principal, the fee is 1.5 percent and the combined quoted cost is 3.5 percent. A zero advertised fee with a four percent exchange-rate margin would be more expensive on that comparison. These are hypothetical prices, not current offers.

Provider comparisons also require identical timing and service specifications. The World Bank’s United States–Mexico page inspected for this report displayed an August 2025 collection window, separate transfer amounts and different access and receiving methods. Those historical observations are not presented here as live recommendations. A comparison should match the principal, funding instrument, settlement currency, payout method and collection date, including whether an offer is promotional. World Bank, United States–Mexico corridor observations.

Travel and time costs can change the household’s preferred service. A nominally cheaper provider requiring a distant collection point may impose transportation expense and lost working time. A faster service may be valuable when a medical payment is due. Account-based delivery may support saving but require an accessible account, identification, connectivity and confidence in resolving errors. These factors should be measured directly rather than assumed from the availability of an app.

A fixed fee creates a tradeoff between sending frequency and total cost. Combining payments can lower fees per dollar, but delay access to resources. A household with little liquidity may prefer more frequent small transfers despite higher unit costs. The relevant welfare comparison includes the costs of waiting or borrowing. There is no single optimal frequency independent of the household’s cash calendar.

Remittances can help a family absorb a local loss of earnings, an illness or a crop failure. Their insurance value depends on the sender having resources when the recipient needs them. A migrant facing unemployment, reduced working hours or higher living costs may be unable to increase support. A shock that weakens employment on both sides of the border creates a different risk from a local event confined to the recipient’s community.

The response can also be nonlinear. A sender may initially draw down savings to maintain transfers, making remittances appear resilient while the household network’s combined buffer declines. After savings are depleted, transfers may fall abruptly. A short stable period does not prove the income source is permanently insulated from employment conditions. For analysis, both sending capacity and the recipient’s need should be considered.

The exchange rate can provide a partial offset. If dollar transfers decline while pesos per dollar rise, peso receipts may fall less or even increase. Domestic inflation can erode that gain. The result depends on the magnitudes and timing, not on an unconditional claim that depreciation protects recipients. A family purchasing imported medicine or equipment may face costs that rise with the same exchange-rate movement.

Different households also have different domestic buffers. An older recipient with limited labor-market options may respond differently from a working household that can increase hours. Existing debt service and health needs can make spending less adjustable. A household with liquid savings may maintain consumption; one with an illiquid home but little cash may not. Wealth, liquidity and current income should therefore be observed separately.

An income-insurance interpretation should be tested against alternative explanations for transfers. A payment can finance a planned house, a family business, an inheritance-related transaction or a temporary expense. An unusually large transfer may reflect a shift in timing rather than a permanent income increase. Regularity and stated purpose can help interpret the pattern, although respondent reports themselves require careful validation.

Alcaraz, Chiquiar and Salcedo’s Banco de México working paper studies the 2008–2009 US recession using a Mexican household panel. It compares children in previously remittance-receiving households with a comparison group and instruments recipient-group membership using historical railway-related distance to the US border. The authors report higher child labor and lower school attendance in their instrumental-variable specifications after the adverse remittance shock. Their simple difference-in-differences estimates are not statistically significant. The distinction matters: the result depends materially on the identification approach. Alcaraz, Chiquiar and Salcedo, working paper 2010-14, methods and results, pages 7–14.

The paper supports taking liquidity-sensitive household adjustments seriously, with the qualifications attached to its setting. It does not provide a coefficient that can be mechanically applied to the 2025 national decline, a current estimate for all recipient households or proof that any particular contemporary payment technology prevents educational losses. This report has inspected the published identification discussion and results but has not replicated the microdata analysis.

The broader methodological lesson is that recipients and nonrecipients are selected groups. Migration networks, household preferences and local labor markets can influence both receipt and outcomes. Comparing their average consumption or schooling can confound those differences with the transfer effect. A historical instrument requires an exclusion argument: the historical determinant must affect the current outcome through the relevant remittance relationship rather than through independent development paths. Controls can address some concerns without proving that every possible channel has been removed.

A current evaluation would need to distinguish a change in transfer amount from the loss of a migrant household member’s other contributions. It would also need to account for local employment, public support and price changes during the same period. A common US shock can affect Mexican exports as well as remittances. Treating all recipient-community deterioration as a remittance effect would ignore that additional exposure.

A household can use a transfer to purchase food, keep a child in school, repay costly debt, repair a home or finance inventory. Each can improve welfare, but their timing and connection to measured production differ. Labeling all consumption as unproductive overlooks health and education. Labeling every asset purchase as productive investment overlooks resale assets, land-price changes and projects with low returns. The relevant mechanism must be specified.

Income is also fungible. A respondent may say that a remittance paid for food while domestic wages financed a machine. The same household might have purchased the food without the transfer and postponed the machine. Asking which source financed which purchase does not necessarily identify the marginal use of the transfer. A causal expenditure response requires a counterfactual household budget, not only an accounting label assigned after the payment arrives.

Consider a hypothetical household with domestic income of 8,000 pesos and remittances of 4,000 per month. It spends 10,000 and saves 2,000. If transfers decline by 1,000 and spending stays unchanged, saving falls to 1,000. The immediate consumption response is zero while the loss of resources is real. The persistent loss reduces planned asset accumulation. A larger loss could require withdrawals and eventually force a spending adjustment after liquid buffers are depleted. A one-month spending observation cannot characterize the full adjustment path.

Alternatively, the household could repay high-cost debt after receiving a temporary transfer. Current consumption might change little while future interest payments fall. That balance-sheet improvement is valuable even though it is not immediately visible as higher retail sales. Business and policy analysis should therefore examine debt, arrears and liquid balances alongside consumption, with care to separate borrowing flows from current income.

At the national level, imported purchases absorb part of the demand impulse, while domestic firms may respond through prices, hours or output. The response depends on capacity and the goods purchased. There is no accounting identity that converts one dollar of remittances into a fixed amount of GDP. Transfers can support domestic demand, but their multiplier is a behavioral estimate requiring its own evidence.

A large state or municipality can receive more remittances because it has more residents or an extensive migrant network. A smaller area can receive less in total while relying more heavily on transfers per resident or relative to household income. These measures answer different questions. Gross receipts identify payment volume; per-capita receipts identify an average intensity; recipient-household shares describe reported budget exposure. A policy ranking should specify which concern it addresses.

Geographic attribution is not automatically residence-based welfare measurement. Collection location, account location and household residence can differ. A recipient may travel to a financial center or use an account associated with another place. Before mapping local receipts onto local income, an analyst should examine how the source assigns geography and whether revisions or payment-channel changes affect that assignment. Fine geographic detail is useful only when the underlying concept remains clear.

Concentration within an area matters as well. If a small group receives most transfers, average receipts can overstate the support available to other residents. Local retailers may nevertheless experience indirect demand from recipients. Those spillovers can benefit workers who receive no remittances, while rent or land-price pressures can create distributional costs. The net effect depends on local supply and who owns the scarce assets.

A hypothetical locality receives one million pesos, of which 600,000 is spent on locally supplied goods and services and 400,000 saved or spent outside the locality. If local businesses use 300,000 of the local purchases to obtain imported or outside inputs, their immediate local gross value added is 300,000 in this deliberately simplified account. Further spending rounds, capacity changes and price responses are excluded. The example shows why gross receipts and local production cannot be equated; it is not a calibrated regional multiplier.

For lenders and businesses, dependable local demand depends on the distribution and persistence of payments as well as the aggregate. Credit extended against an unusually high transfer month can create repayment pressure when timing normalizes. Evidence on recurring receipts, household obligations and alternative income is more useful than extrapolating a national growth rate into every local customer’s budget.

A resource shock can be manageable over a year and still create an immediate payment failure. The difference is the timing of receipts and obligations. Consider a hypothetical household with monthly domestic income of 8,000 pesos and net remittance receipts of 4,000. Its recurring essential commitments are 9,000, and it holds liquid savings of 3,000. Before other discretionary purchases, the household has a monthly resource margin of 3,000. This is an assumed budget, not an estimate for a typical Mexican family.

Suppose the remittance stops for two months while domestic income and essential commitments remain unchanged. The household’s recurring deficit becomes 1,000 per month. Its liquid savings can cover that deficit for three months if no other spending or emergency occurs. The two-month interruption reduces savings from 3,000 to 1,000. Spending on essentials can remain stable, but the household finishes with substantially less capacity to absorb another loss. A study using only essential consumption would miss that decline in resilience.

Now add a medical expense of 2,000 during the second month. Available savings no longer cover both the accumulated recurring deficit and the medical expense: total financing needs are 4,000 against an initial liquid buffer of 3,000. The shortfall is 1,000. Access to an affordable loan, a public service or support from another relative changes the adjustment. Selling an asset can also produce cash, but the future consequences depend on whether the asset is replaceable or essential to earning income.

A later recovery of transfers does not automatically reverse every consequence. Deferred treatment, interrupted schooling, a costly loan or a distressed asset sale can create persistent costs. Alternatively, a temporary reduction in discretionary purchases may have limited long-run effects. The outcome depends on the household’s commitments and adjustment options. It should be observed rather than inferred from the duration of the aggregate national decline.

The stress test also distinguishes exchange-rate exposure from employment exposure. A stable dollar payment can produce fewer pesos after appreciation; the entire payment can disappear after a sender loses income. Both reduce resources, but their probability, duration and correlation with other household income differ. Adding these risks as though they were independent can understate joint exposure. Assuming they always occur together can overstate it. Scenario analysis should show the combinations being considered and avoid assigning unsupported probabilities.

When recipients spend more on goods whose local supply is slow to expand, part of the demand increase can appear as higher prices. Housing repairs may compete for a limited number of skilled workers; a supply-constrained rental market can pass some additional purchasing power to landlords. Other goods can be obtained from outside the locality at relatively stable prices, allowing quantities to respond more readily. A general claim that remittances either cause inflation or only increase output ignores these differences.

The distributional interpretation depends on ownership and household circumstances. A recipient who owns a home may benefit from improved property values while a tenant faces higher rent. A nonrecipient construction worker may gain earnings from stronger demand. A nonrecipient living on a fixed income may face higher costs without an offsetting income increase. Aggregate local receipts cannot determine which of these effects dominates for a particular group.

Empirical analysis would distinguish a change in the quantity and quality of purchased housing from a pure price increase. A more expensive dwelling may be larger or better located. Price comparisons that fail to control for those attributes can misstate inflation. Conversely, treating every higher purchase price as improved quality would conceal real affordability pressure. Local price evidence and household expenditure quantities are needed to separate these mechanisms.

A credible assessment would connect the household receipt record with settlement terms and a household budget observed over time. The required variables include gross sending budget, fee convention, applied exchange rate, net pesos received, collection cost, receipt frequency and the number of beneficiaries sharing resources. Other income, liquid assets, debt service and relevant expenditure prices are needed to interpret the transfer’s effect. Consent and confidentiality are necessary conditions for such linked research, rather than optional additions to an administrative dataset.

The first stage would reconcile amounts without making a causal claim. It would compare recorded net receipts with respondent reports, identify duplicated household transfers and distinguish seasonal timing from persistent loss. Survey weights and attrition would need attention. A household that disappears from a panel after losing support may be precisely the case most relevant to resilience, so a balanced-panel result should be accompanied by an assessment of missing observations.

A second stage would identify the source of variation. Changes in sender employment, payment charges, settlement rates and recipient need imply different treatments. A provider’s fee reduction may attract households with different characteristics, while a recipient’s urgent expense can itself cause a larger transfer. An observed association between more remittances and higher spending is consequently insufficient to identify the effect of an externally determined resource gain.

Where a policy or pricing change creates a credible comparison, analysis should test preexisting trends, exposure definitions and alternative channels. Randomized information about prices can identify search or switching behavior without necessarily identifying a long-run income effect. An intervention that changes account access can affect saving and credit independently of transfer costs. Distinguishing these mechanisms yields more useful conclusions than assigning all changes to financial inclusion as a single outcome.

The appropriate outputs would include real resources per person, consumption stability, liquid-buffer depletion, debt stress and relevant human-capital outcomes. Heterogeneity by income, age, location and household composition should be reported when the sample supports it. This report supplies an accounting framework and source-based cautions for that work; it does not claim to have collected a new recipient panel or measured those causal effects.

The next dollar-flow release should first be placed in its own time frame and revision vintage. Monthly, cumulative and rolling-year measures can move differently. Transaction counts and average values explain the reported aggregate mechanically, but do not establish household participation. Exchange-rate and price indices can then describe a standardized real-peso trend, with their limitations stated. Household evidence is needed to establish distribution, exposure and the adjustment of spending and balance sheets.

A careful interpretation can recognize the importance of remittances without promising that their growth guarantees rising living standards. It can also identify a decline without assuming an equal immediate fall in consumption or a known number of households losing support. The economically decisive question is how much usable purchasing power reaches particular families, how reliable that support remains and what other resources they have when it changes.